SYSTEMS AND METHODS FOR OPTIMIZING PRODUCT FEEDS FOR PRICE COMPARSION SHOPPING WEBSITES

Application No. 18/460,336

• Filed: Sep 01, 2023
• Examiner: WEINER, ARIELLE E
• Art Unit: 3689
• Tech Center: 3600 — Transportation & Electronic Commerce
• Assignee: Coupang Corp.
• OA Round: 3 (Non-Final)

Office Action

DETAILED ACTION
This action is in reply to the Amendments filed on 06/24/2026.
Claims 1-20 are rejected.
Claims 1-20 are currently pending and have been examined.

Response to Amendment
Applicant’s amendment, filed 06/24/2026, has been entered. Claims 1, 11, and 16-20 have been amended.

Claim Rejections under 35 USC § 112(b)
Some of the USC § 112(b) claim rejections from the prior Office Action have been withdrawn pursuant Applicant’s amendments. The remaining rejections are detailed below.

Information Disclosure Statement
Information Disclosure Statement received 04/22/2026 has been reviewed and considered.
Continued Examination Under 37 CFR 1.114
A request for continued examination under 37 CFR 1.114, including the fee set forth in 37 CFR 1.17(e), was filed in this application after final rejection. Since this application is eligible for continued examination under 37 CFR 1.114, and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114. Applicant's submission filed on 06/24/2026 has been entered.

Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .

Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claims 1-20 are rejected under 35 U.S.C. 101 because the claimed invention is directed to a judicial exception (i.e., law of nature, a natural phenomenon, or an abstract idea) without significantly more.
Under Step 1 of the Subject Matter Eligibility Test for Products and Processes, the claims must be directed to one of the four statutory categories (see MPEP 2106.03). All the claims are directed to one of the four statutory categories (YES).

Under Step 2A of the Subject Matter Eligibility Test, it is determined whether the claims are directed to a judicially recognized exception (see MPEP 2106.04). Step 2A is a two-prong inquiry.

Under Prong 1, it is determined whether the claim recites a judicial exception (YES). Taking Claim 20 as representative, the claim recites limitations that fall within the certain methods of organizing human activity groupings of abstract ideas, including:
-A non-transitory computer readable medium including instructions that are executable by one or more processors to cause a system to perform a method, the method comprising:
-obtaining historical product data corresponding to a first product set, the first product set comprising one or more products for display on a webpage associated with the system;
-associating each product in the first set of products with a product [data] feed key, the product [data] feed key comprising a product identifier and a price;
-determining a capacity constraint associated with a second system configured to [that] display[s] information associated with the one or more products and receiving the capacity constraint from the second system;
-wherein the capacity constraint comprises a numerical limit based on a memory constraint of the second system;
-generating a gradient boosting machine regression model configured to generate a predicted amount of interactions corresponding to at least one product in the first product set;
-training [updating], at training intervals based on the product feed, the gradient boosting machine regression model by:
-generating aggregated product data based on the historical product data;
-inputting the aggregated product data into the gradient boosting machine regression model;
-obtaining from the second system: the product [data] feed; and a product feed click count;
-providing at least one of the product [data] feed or the product feed click count to the gradient boosting machine learning regression model; and
-updating one or more weights in the gradient boosting machine regression model based on the at least one of the product [data] feed or the product feed click count;
-optimizing the trained gradient boosting machine regression model by applying Bayesian optimization to one or more hyperparameters associated with the trained gradient boosting machine regression model;
-generating, with the optimized gradient boosting machine regression model, the predicted amount of interactions;
-wherein the predicted amount of interactions corresponds to a future time interval;
-selecting, based on the predicted amount of interactions, a second product set, wherein the second product set includes a subset of the one or more products from the first product set based on the capacity constraint;
-mapping each product [data] feed key of the second set of products to a product in the second system; and
-presenting the second product set to the second system for display during the future time interval
The above limitations recite the concept of utilizing historical data of a first set of products, a capacity constraint, and a predicted amount of future product interactions to select a subset of the first set of products. The above limitations fall within the “Certain Methods of Organizing Human Activity” groupings of abstract ideas, enumerated in MPEP 2106.04(a).
Certain methods of organizing human activity include:
fundamental economic principles or practices (including hedging, insurance, and mitigating risk)
commercial or legal interactions (including agreements in the form of contracts; legal obligations; advertising, marketing or sales activities or behaviors; and business relations)
managing personal behavior or relationships or interactions between people (including social activities, teaching, and following rules or instructions)

The limitations of wherein the capacity constraint comprises a numerical limit based on a memory constraint of the second system; wherein the predicted amount of interactions corresponds to a future time interval; selecting, based on the predicted amount of interactions, a second product set, wherein the second product set includes a subset of the one or more products from the first product set based on the capacity constraint; and presenting the second product set to the second system for display during the future time interval are processes that, under their broadest reasonable interpretation, cover a commercial interaction. For example, “generating” “selecting,” and “presenting” in the context of this claim encompass advertising, and marketing or sales activities.

Similarly, the limitations of a non-transitory computer readable medium including instructions that are executable by one or more processors to cause a system to perform a method, the method comprising: obtaining historical product data corresponding to a first product set, the first product set comprising one or more products for display on a webpage associated with the system; associating each product in the first set of products with a product [data] feed key, the product [data] feed key comprising a product identifier and a price; determining a capacity constraint associated with a second system configured to [that] display[s] information associated with the one or more products and receiving the capacity constraint from the second system; generating a gradient boosting machine regression model configured to generate a predicted amount of interactions corresponding to at least one product in the first product set; training [updating], at training intervals based on the product feed, the gradient boosting machine regression model by: inputting the aggregated product data into the gradient boosting machine regression model; obtaining from the second system: the product [data] feed; and a product feed click count; providing at least one of the product [data] feed or the product feed click count to the gradient boosting machine learning regression model; and updating one or more weights in the gradient boosting machine regression model based on the at least one of the product [data] feed or the product feed click count; optimizing the trained gradient boosting machine regression model by applying Bayesian optimization to one or more hyperparameters associated with the trained gradient boosting machine regression model; generating, with the optimized gradient boosting machine regression model, the predicted amount of interactions; and mapping each product [data] feed key of the second set of products to a product in the second system are processes that, under their broadest reasonable interpretation, cover a commercial interaction. That is, other than reciting that a non-transitory computer readable medium including instructions that are executable by one or more processors to causes the system to perform a method, that product data key is a product data feed key, that product data is a product data feed, that the displaying is on a webpage, the second system is ‘configured to’ display, that the model is a gradient boosting machine regression model, that the model is trained, that the product click count is a product feed click count, that the model is a trained gradient boosting machine regression model, and that the model is an optimized gradient boosting machine regression model, nothing in the claim element precludes the step from practically being performed by people. For example, but for the “non-transitory computer readable medium including instructions that are executable by one or more processors,” “feed,” “trained,” “webpage,” “configured to,” “gradient boosting machine regression model,” “training” “trained gradient boosting machine regression model,” and “optimized gradient boosting machine regression model” language, “perform,” “obtaining,” “associating,” “determining,” “generating,” “training,” “inputting,” “obtaining,” “providing,” “updating,” “optimizing,” “generating,” and “mapping” in the context of this claim encompasses advertising, and marketing or sales activities.

Under Prong 2, it is determined whether the claim recites additional elements that integrate the exception into a practical application of the exception. This judicial exception is not integrated into a practical application (NO).
-A non-transitory computer readable medium including instructions that are executable by one or more processors to cause a system to perform a method, the method comprising:
-obtaining historical product data corresponding to a first product set, the first product set comprising one or more products for display on a webpage associated with the system;
-associating each product in the first set of products with a product feed key, the product feed key comprising a product identifier and a price;
-determining a capacity constraint associated with a second system configured to display information associated with the one or more products and receiving the capacity constraint from the second system;
-wherein the capacity constraint comprises a numerical limit based on a memory constraint of the second system;
-generating a gradient boosting machine regression model configured to generate a predicted amount of interactions corresponding to at least one product in the first product set;
-training, at training intervals based on the product feed, the gradient boosting machine regression model by:
-generating aggregated product data based on the historical product data;
-inputting the aggregated product data into the gradient boosting machine regression model;
-obtaining from the second system: the product feed; and a product feed click count;
-providing at least one of the product feed or the product feed click count to the gradient boosting machine learning regression model; and
-updating one or more weights in the gradient boosting machine regression model based on the at least one of the product feed or the product feed click count;
-optimizing the trained gradient boosting machine regression model by applying Bayesian optimization to one or more hyperparameters associated with the trained gradient boosting machine regression model;
-generating, with the optimized gradient boosting machine regression model, the predicted amount of interactions;
-wherein the predicted amount of interactions corresponds to a future time interval;
-selecting, based on the predicted amount of interactions, a second product set, wherein the second product set includes a subset of the one or more products from the first product set based on the capacity constraint;
-mapping each product feed key of the second set of products to a product in the second system; and
-presenting the second product set to the second system for display during the future time interval
The additional elements of claim 20 are recited at a high level of generality (i.e. as generic computing hardware) such that they amount to nothing more than mere instructions to implement or apply the abstract idea on a generic computing hardware (or, merely use a computer as a tool to perform an abstract idea) as supported by paragraph [0102] of Applicant’s specification – “Various programs or program modules can be created using any of the techniques known to one skilled in the art or can be designed in connection with existing software.” Specifically, the additional element of non-transitory computer readable medium including instructions that are executable by one or more processors, a product feed key, a product feed, trained, webpage, configured to, gradient boosting machine regression model, training trained gradient boosting machine regression model,” and “optimized gradient boosting machine regression model is recited at a high-level of generality (i.e. as a generic processor performing the generic computer functions of obtaining data, associating data, determining data, generating data, training data, inputting data, providing data, updating data, optimizing data, selecting data, mapping data, and presenting data)

such that they amount do no more than mere instructions to apply the exception using generic computer components. Accordingly, these additional elements do not integrate the abstract idea into a practical application because they do not impose any meaningful limits on practicing the abstract idea. The claim is directed to an abstract idea. Further, the additional elements do no more than generally link the use of the judicial exception to a particular technological environment or field of use (such as computers or computing networks). Employing well-known computer functions to execute an abstract idea, even when limiting the use of the idea to one particular environment, does not integrate the exception into a practical application.

Additionally, the additional elements are insufficient to integrate the abstract idea into a practical application because the claim fails to i) reflect an improvement in the functioning of a computer or an improvement to another technology or technical field, ii) apply the judicial exception with, or use the judicial exception in conjunction with, a particular machine or manufacture that is integral to the claim, iii) effect a transformation or reduction of a particular article to a different state or thing, or iv) apply or use the judicial exception in some other meaningful way beyond generally linking the use of the judicial exception to a particular technological environment.
Accordingly, the judicial exception is not integrated into a practical application.

Under Step 2B, it is determined whether the claims recite additional elements that amount to significantly more than the judicial exception. The claims of the present application do not include additional elements that are sufficient to amount to significantly more than the judicial exception (NO).

In the case of claim 20, taken individually or as a whole, the additional elements of claim 9 do not provide an inventive concept. As discussed above under step 2A (prong 2) with respect to the integration of the abstract idea into a practical application, the additional elements used to perform the claimed functions amount to no more than a general link to a technological environment.

Even considered as an ordered combination (as a whole), the additional elements do not add anything significantly more than when considered individually.

Claim 1 is a system reciting similar functions as claim 20. Examiner notes that claim 1 recites the additional elements of at least one memory storing instructions, at least one processor configured to execute the instructions, a product feed, a product feed key, machine learning models that are trained, training, and a webpage, however, claim 1 does not qualify as eligible subject matter for similar reasons as claim 20 indicated above.

Claim 11 is a method reciting similar functions as claim 20. Examiner notes that claim 11 recites the additional elements of a product feed, a product feed key, machine learning models that are trained, training, and a webpage, however, claim 11 does not qualify as eligible subject matter for similar reasons as claim 20 indicated above.

Therefore, claims 1, 11, and 20 do not provide an inventive concept and do not qualify as eligible subject matter.

Dependent claims 2-10 and 12-19, when analyzed as a whole, are held to be patent ineligible under 35 U.S.C. § 101 because they do not add “significantly more” to the abstract idea. More specifically, dependent claims 2-10 and 12-19 further fall within the “Certain Methods of Organizing Human Activity” grouping of abstract ideas in that they recite commercial interactions. Dependent claim 10 does not recite any farther additional elements, and as such are not indicative of integration into a practical application for at least similar reasons discussed above. Dependent claims 2-9 and 12-19 recite the additional elements of training of the machine learning model, a database, the machine learning model, a product feed, a product feed click count, the website, memory consumption, and a product feed key, but similar to the analysis under prong two of Step 2A these additional elements are used as a tool to perform the abstract idea. As such, under prong two of Step 2A, claims 2-10 and 12-19 are not indicative of integration into a practical application for at least similar reasons as discussed above. Thus, dependent claims 2-10 and 12-19 are “directed to” an abstract idea. Next, under Step 2B, similar to the analysis of claims 1, 11, and 20, dependent claims 2-10 and 12-19 when analyzed individually and as an ordered combination, merely further define the commonplace business method (i.e. utilizing historical data of a first set of products, a capacity constraint, and a predicted amount of future product interactions to select a subset of the first set of products) being applied on a general-purpose computer and, therefore, do not amount to significantly more than the abstract idea itself. Accordingly, the Examiner concludes that there are no meaningful limitations in the claims that transform the judicial exception into a patent eligible application such that the claims amount to significantly more than the judicial exception itself. The analysis above applies to all statutory categories of invention.

Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
This application currently names joint inventors. In considering patentability of the claims the examiner presumes that the subject matter of the various claims was commonly owned as of the effective filing date of the claimed invention(s) absent any evidence to the contrary. Applicant is advised of the obligation under 37 CFR 1.56 to point out the inventor and effective filing dates of each claim that was not commonly owned as of the effective filing date of the later invention in order for the examiner to consider the applicability of 35 U.S.C. 102(b)(2)(C) for any potential 35 U.S.C. 102(a)(2) prior art against the later invention.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.

The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
Claims 1-20 are rejected under 35 U.S.C. 103 as being unpatentable over Hendlin et al. (US 2018/0053244 A1), hereinafter Hendlin , in view of Lakhani et al. (US 2024/0257165 A1), hereinafter Lakhani, further view of Abuomar et al. (US 10,133,759 B1), hereinafter Abuomar .
Regarding claim 1, Hendlin discloses a system comprising:
-at least one memory storing instructions (Hendlin, see at least: “processor 1002 includes hardware for executing instructions, such as those making up a computer program. As an example and not by way of limitation, to execute instructions, processor 1002 may retrieve (or fetch) the instructions from an internal register, an internal cache, memory 1004, or storage device 1006 and decode and execute them [i.e. at least one memory storing instructions]” [0218]);
-at least one processor configured to execute the instructions to perform operations for optimizing a product feed based on a model (Hendlin, see at least: “The components 702-716 and their corresponding elements can comprise software, hardware, or both. For example, the components 702-716 and their corresponding elements can comprise one or more instructions stored on a computer-readable storage medium and executable by processors of one or more computing devices [i.e. at least one processor configured to execute the instructions to perform operations]” [0195] and “the ranking algorithm can apply criteria directed to product performance (e.g., product sales, product clicks, product views, product discussion from social networking system data [i.e. optimizing a product feed based on a model] or a product feed)” [0132]), the operations comprising:
-obtaining historical product data corresponding to a first product set, the first product set comprising one or more products for display on a webpage associated with the system (Hendlin, see at least: “the step 120 can also include obtaining a product feed from a website that includes a merchant's products or services [i.e. the first product set comprising one or more products for display on a webpage associated with the system]. In particular, the social networking system 104 can obtain a product feed from a website by parsing the website and identifying product information. For example, in one or more embodiments, the social networking system 104 parses a website and identifies product names, product images and/or video, product prices, product sale status (e.g., products on sale), or product stock status” [0062] and “the ranking algorithm can apply criteria directed to product performance (e.g., product sales, product clicks, product views, product discussion from social networking system data or a product feed) [i.e. obtaining historical product data corresponding to a first product set] … product information (e.g., a product order on an existing website, product emphasis, sale status, stock status from a product feed)” [0132] and “the product feed 450 can include updating information regarding product sales, product popularity, or other metrics (e.g., clicks on other websites, online purchases, or survey data). The ranking algorithm 460 can generate a score based on the extent to which a particular product corresponds to increased sales, popularity, clicks, purchases, or other metrics from the product feed 450” [0136] “the ranking algorithm can utilize criteria directed to location, user history (e.g., a user's previous views, purchases, or clicks) [i.e. obtaining historical product data corresponding to a first product set]” [0138]);
-associating each product in the first set of products with a product feed key, the product feed key comprising a product identifier and a price (Hendlin, see at least: “the term “product feed” refers to a digital item defining product information. In particular, the term “product feed” includes an updating digital item that defines product information corresponding to products or services of a merchant over time. A product feed can take a variety of forms, for example, a product feed can comprise a database, spreadsheet, text file, or other electronic file [i.e. associating each product in the first set of products with a product feed key]. For instance, a merchant can maintain an updating database of products or services and provide the updating database as a product feed. In addition, the product feed can comprise a URL to a website. For example, a merchant can maintain an updating website that describes products or services and provide the website as a product feed” [0045] and “the digital merchant content system obtains a product feed comprising an updating database [i.e. the product feed key] with product information from a merchant (e.g., an updating database of product names [i.e. comprising a product identifier], product styles, product groups, product sales, product prices [i.e. comprising a price], and/or other product attributes)” [0033]);
-determining a capacity constraint associated with a second system configured to display information associated with the one or more products (Hendlin, see at least: “the ranking algorithm 460 can rank the products 454a-454n by score and then select a number of products corresponding to a number of unpopulated elements in an unpopulated product display layout (i.e., the highest ranked products needed to fill the unpopulated elements) [i.e. determining a capacity constraint associated with a second system configured to display information associated with the one or more products]” [0147] and “the social networking system 104 can also perform the step 138 of generating custom merchant content interfaces [i.e. associated with a second system configured to display information associated with the one or more products]. In particular, the social networking system 104 can generate a custom merchant content interface based on a custom merchant content template, a product feed, and/or social networking system data” [0071] Examiner notes that the number of unpopulated elements is the capacity constraint);
-generating an amount of interactions corresponding to at least one product in the first product set, wherein the generating is based on on the historical product data (Hendlin, see at least: “The digital merchant content system 100 can also determine metrics of product performance via the product feed 450. For example, the product feed 450 can include updating information regarding product sales, product popularity, or other metrics (e.g., clicks on other websites, online purchases, or survey data). The ranking algorithm 460 can generate a score based on the extent to which a particular product corresponds to increased sales, popularity, clicks, purchases, or other metrics from the product feed 450 [i.e. generating an amount of interactions corresponding to at least one product in the first product set]” [0136] and “the ranking algorithm can apply criteria directed to product performance (e.g., product sales, product clicks, product views, product discussion from social networking system data or a product feed) [i.e. wherein the generating is based on the historical product data], user characteristics (e.g., demographic information, location information, or user history), product information (e.g., a product order on an existing website, product emphasis, sale status, stock status from a product feed), or a combination of such criteria” [0132]);
-selecting, based on the amount of interactions, a second product set, wherein the second product set includes a subset of the products from the first product set based on the capacity constraint (Hendlin, see at least: “the digital merchant content system 100 provides the product feed 450, the merchant preferences 456, and the social networking system data 458 to a ranking algorithm 460 and the ranking algorithm 460 generates selected products 462. [i.e. selecting a second product set, wherein the second product set includes a subset of the products from the first product] In particular, the ranking algorithm 460 generates the selected products 462 to utilize in populating unpopulated elements in the custom merchant content template 420 [i.e. based on the capacity constraint]. In one or more embodiments, the ranking algorithm 460 calculates a score corresponding to each product 454a-454n. In particular, the ranking algorithm 460 applies criteria to each of the products 454a-454n and calculates a score corresponding to each of the products 454a-454n based on whether (and/or to what extent) each of the products 454a-454n satisfy the criteria” [0131] and “the ranking algorithm can apply criteria directed to product performance (e.g., product sales, product clicks, product views, product discussion from social networking system data or a product feed) [i.e. based on the amount of interactions], user characteristics (e.g., demographic information, location information, or user history), product information (e.g., a product order on an existing website, product emphasis, sale status, stock status from a product feed), or a combination of such criteria” [0132] and “The digital merchant content system 100 can also determine metrics of product performance via the product feed 450. For example, the product feed 450 can include updating information regarding product sales, product popularity, or other metrics (e.g., clicks on other websites, online purchases, or survey data). The ranking algorithm 460 can generate a score based on the extent to which a particular product corresponds to increased sales, popularity, clicks, purchases, or other metrics from the product feed 450 [i.e. based on the amount of interactions]” [0136] Examiner notes that the number of unpopulated elements is the capacity constraint);
-mapping each product feed key of the second set of products to a product in the second system (Hendlin, see at least: “the merchant, via the merchant device 102, can provide the social networking system 104 with an indication of a digital location of an updating database that outlines product information corresponding to the merchant's products. The social networking system 104 can access the product feed by obtaining the updating database from the location identified by the merchant [i.e. mapping each product feed key of the second set of products to a product in the second system]” [0061] and “the social networking system 104 can also perform the step 122 of storing the product feed (e.g., storing product information received from the product feed) [i.e. to a product in the second system]. In particular, the step 122 can include storing product information corresponding to a product feed at the social networking system 104 so that the social networking system 104 can access product information. For example, in one or more embodiments, the step 122 comprises storing a product feed and associating a product feed with a corresponding merchant (e.g., associating the product feed with an account of the merchant) [i.e. mapping each product feed key of the second set of products to a product in the second system]” [0065] and “the digital merchant content system can also identify a product feed. For instance, in one or more embodiments, the digital merchant content system obtains a product feed comprising an updating database with product information from a merchant (e.g., an updating database of product names, product styles, product groups, product sales, product prices, and/or other product attributes)” [0033]);
-presenting the second product set to the second system to update the product feed (Hendlin, see at least: “upon generating a score for the products 454a-454n, the ranking algorithm 460 can identify the selected products 462. In particular, in one or more embodiments, the ranking algorithm 460 identifies the selected products 462 by comparing scores corresponding to the products 454-454n. For example, the ranking algorithm 460 can rank the products 454a-454n by score and then select a number of products corresponding to a number of unpopulated elements in an unpopulated product display layout (i.e., the highest ranked products needed to fill the unpopulated elements) [i.e. presenting the second product set]” [0147] and “a digital merchant content system that provides one or more custom merchant content interfaces within a system and/or application that is separate from a merchant (e.g., a social networking system and/or application) [i.e. to the second system]. In particular, in one or more embodiments, the digital merchant content system generates custom merchant content interfaces that provide targeted, up-to-date, merchant content within a social networking system [i.e. to update the product feed]” [0026]); and
-causing the second system to display the updated the product feed (Hendlin, see at least: “a digital merchant content system that provides one or more custom merchant content interfaces within a system and/or application that is separate from a merchant (e.g., a social networking system and/or application). In particular, in one or more embodiments, the digital merchant content system generates custom merchant content interfaces that provide targeted, up-to-date, merchant content within a social networking system [i.e. causing the second system to display the updated the product feed]” [0026] and Fig. 4D).

Hendlin does not explicitly disclose the model being a machine learning model; generating, with a machine learning model, a predicted amount of interactions corresponding to at least one product, wherein the machine learning model is trained on the historical product data at training intervals based on the product feed; and the amount of interactions being a predicted amount of interactions.
Lakhani, however, teaches determining relevant products to provide (i.e. [0003]), including the known technique of a machine learning model (Lakhani, see at least: “a machine learning model is trained based on data of shared items that are being sold both online and in-store by a retailer [i.e. a machine learning model]. For a physical store of the retailer, inference items are determined from items being sold online but not in-store. An estimated demand is computed for each inference item to be offered for sale in the physical store in a future time period, based on the trained machine learning model and online data of the inference item. Based on the estimated demands for the inference items, recommended assortment data is generated for the physical store in the future time period” [0025]);
the known technique of generating, with a machine learning model, a predicted amount of interactions corresponding to at least one product, wherein the machine learning model is trained on the historical product data at training intervals based on the product feed (Lakhani, see at least: “the demand estimation computing device 102 may execute one or more models (e.g., programs or algorithms), such as a machine learning model, deep learning model, statistical model, etc., to compute estimated in-store demands for items, which are currently sold online but not in-store, when these items are offered for sale in the store 109 during a future time period, e.g. next month or next year [i.e. generating a predicted amount of interactions corresponding to at least one product]” [0034] and “the demand estimation computing device 102 may compute an estimated in-store demand for an item based on a machine learning model that is trained based on shared items sold both online and in-store … the in-store demand for the item may be computed also based on: store traffic data of an item category including the item in the store; online price data for the item; online historical demand data for the item [i.e. with a machine learning model, wherein the machine learning model is trained on the historical product data]” [0035] and “the demand estimation computing device 102 generates and/or updates different models for estimating in-store demand based on online data [i.e. at training intervals based on the product feed]. The models, when executed by the demand estimation computing device 102, allow the demand estimation computing device 102 to determine in-store demands and generate recommended assortment data for each store to refresh assortment for a future time period” [0037] and “the models may be trained following the process 500 in FIG. 5 every quarter or every year; and the assortment refresh may be automatically and periodically performed following the process 800 in FIG. 8, every month, every quarter or every year [i.e. at training intervals based on the product feed]” [0101]); and
the known technique of a predicted amount of interactions (Lakhani, see at least: “the demand estimation computing device 102 may execute one or more models (e.g., programs or algorithms), such as a machine learning model, deep learning model, statistical model, etc., to compute estimated in-store demands for items, which are currently sold online but not in-store, when these items are offered for sale in the store 109 during a future time period, e.g. next month or next year [i.e. a predicted amount of interactions]” [0034]). These known techniques are applicable to the system of Hendlin as they both share characteristics and capabilities, namely, they are directed to determining relevant products to provide.
It would have been recognized that applying the known techniques of a machine learning model; generating, with a machine learning model, a predicted amount of interactions corresponding to at least one product, wherein the machine learning model is trained on the historical product data at training intervals based on the product feed; and a predicted amount of interactions, as taught by Lakhani, to the teachings of Hendlin would have yielded predictable results because the level of ordinary skill in the art demonstrated by the references applied shows the ability to incorporate such references into similar systems. Further, adding the modifications of a machine learning model; generating, with a machine learning model, a predicted amount of interactions corresponding to at least one product, wherein the machine learning model is trained on the historical product data at training intervals based on the product feed; and a predicted amount of interactions, as taught by Lakhani, into the system of Hendlin would have been recognized by those of ordinary skill in the art as resulting in an improved system that would recommend an assortment of items based on estimated demands (Lakhani, [0021]).

Hendlin in view of Lakhani does not explicitly teach receiving the capacity constraint from the second system, wherein the capacity constraint is based on a memory constraint of the second system.
Abuomar, however, teaches rules associated with a capacity (i.e. Col. 10 Ln. 9-16), including the known technique of receiving the capacity constraint from the second system, wherein the capacity constraint is based on a memory constraint of the second system (Abuomar, see at least: “the rule data 122 may not include rules associated with a particular entity identifier 110 … The other rules may output to the entity 108 to solicit user input confirming or rejecting the rules [i.e. receiving the capacity constraint from the second system]” Col. 6 Ln. 53-59 and “A rule module 120 associated with the processing server 106 may determine one or more rules associated with the received data objects 104. In some implementations, the rule module 120 may access rule data 122 [i.e. from the second system]” Col. 6 Ln. 23-26 and “FIG. 3 depicts a scenario 300 illustrating a method for determining one or more rules 210 associated with a data object 104 and providing the data object 104 to a data store 102 indicated by the outcome of the rule(s) 210. At 302, a data object 104 may be received by a persistence server 106. In some implementations, the data object 104 may be provided by an entity 108, such as a service, a user, or another type of process or device. In other implementations, the data object 104 may be accessed by the persistence server 106. For example, the data object 104 may be stored in association with the persistence server 106 or in association with a data store 102 or computing device in communication with the persistence server 106. The data object 104 may include data object characteristics 204. For example, the data object characteristics 204 may include physical characteristics of the data object 104, such as a size of the data object (e.g., 2.3 MB) [i.e. wherein the capacity constraint is based on a memory constraint of the second system]” Col. 9 Ln. 15-31 and “FIG. 3 depicts an example rule 210 that reads: “If (Size(DataObject)>500 MB OR SalesRank(DataObject)>50) {Store in DataStore1} else {Store in DataStore2}”. The example rule 210 includes three expressions 214 and two threshold values 216. The first expression 214 describes a relationship between a data object characteristic 204 (the size of the data object 204) and a threshold value 216 (500 MB) [i.e. wherein the capacity constraint is based on a memory constraint of the second system]” Col. 10 Ln. 9-16 and “Each data object 104 may be stored in a particular data store 102 based on the characteristics of the data object 104 and on the data store characteristics 114 associated with the data stores 102. Data store characteristics 114 may include configurations associated with an associated data store 102, such as the access speed associated with stored data objects 104, a financial cost associated with using or accessing the data store 102, a computational cost associated with use or access of the data store 102, security features associated with the data store 102, storage capacity of the data store 102 [i.e. of the second system], restrictions regarding sizes or types of data objects 104 that may be stored in the data store 102, and so forth” Col. 5 Ln. 42-53 Examiner notes that entities 108 are the system and the persistence server 106 is the second system). This known technique is applicable to the system of Hendlin in view of Lakhani as they both share characteristics and capabilities, namely, they are directed to rules associated with a capacity.
It would have been recognized that applying the known techniques of receiving the capacity constraint from the second system, wherein the capacity constraint is based on a memory constraint of the second system, as taught by Abuomar, to the teachings of Hendlin in view of Lakhani would have yielded predictable results because the level of ordinary skill in the art demonstrated by the references applied shows the ability to incorporate such references into similar systems. Further, adding the modification of receiving the capacity constraint from the second system, wherein the capacity constraint is based on a memory constraint of the second system, as taught by Abuomar, into the system of Hendlin in view of Lakhani would have been recognized by those of ordinary skill in the art as resulting in an improved system that would determine rules associated with the received data objects (Abuomar, Col. 6 Ln. 23-26).

Regarding claim 2, the combination of Hendlin/Lakhani/Abuomar teach the system of claim 1. Hendlin further discloses:
-wherein updating of the model (Hendlin, see at least: “the digital merchant content system 100 can check for updates utilizing a certain schedule (e.g., once a minute, twice an hour, or once a day) [i.e. updating on a daily basis]” [0115] and “The digital merchant content system 100 can also determine metrics of product performance via the product feed 450. For example, the product feed 450 can include updating information regarding product sales, product popularity, or other metrics (e.g., clicks on other websites, online purchases, or survey data). The ranking algorithm 460 can generate a score based on the extent to which a particular product corresponds to increased sales, popularity, clicks, purchases, or other metrics from the product feed 450 [i.e. updating the model]” [0136] and “the digital merchant content system 100 can receive an update to the product feed 450 indicating that a first product is out of stock. Similarly, the digital merchant content system 100 can receive additional social networking system data 458 indicating that a second product is generating increased sales. In response, the digital merchant content system 100 (e.g., via the ranking algorithm 460) can modify the product display layout 442 [i.e. updating the model] (e.g., to remove the first product that is out of stock and to add the second product that is generating increased sales)” [0154]) comprises:
-generating, on a daily basis, aggregated product data based on the historical product data (Hendlin, see at least: “the digital merchant content system 100 can check for updates utilizing a certain schedule (e.g., once a minute, twice an hour, or once a day) [i.e. generating, on a daily basis,]” [0115] and “The digital merchant content system 100 can also determine metrics of product performance via the product feed 450. For example, the product feed 450 can include updating information regarding product sales, product popularity, or other metrics (e.g., clicks on other websites, online purchases, or survey data). The ranking algorithm 460 can generate a score based on the extent to which a particular product corresponds to increased sales, popularity, clicks, purchases, or other metrics from the product feed 450 [i.e. aggregated product data based on the historical product data]” [0136]), by:
-inputting the aggregated product data into the model daily (Hendlin, see at least: “The digital merchant content system 100 can also determine metrics of product performance via the product feed 450. For example, the product feed 450 can include updating information regarding product sales, product popularity, or other metrics (e.g., clicks on other websites, online purchases, or survey data). The ranking algorithm 460 can generate a score based on the extent to which a particular product corresponds to increased sales, popularity, clicks, purchases, or other metrics from the product feed 450 [i.e. inputting the aggregated product data into the model]” [0136] and “the database 300 can include media content items corresponding to products (e.g., a location of product image files or product video files), product specifications, product descriptions, product locations, product shipping items, product timing information (e.g., information indicating a time corresponding to seasonal products), product sale information (e.g., sale volume or sale rank compared to other products) [i.e. the aggregated product data]” [0114] and “the digital merchant content system 100 can check for updates utilizing a certain schedule (e.g., once a minute, twice an hour, or once a day) [i.e. daily]” [0115]).

Hendlin does not explicitly disclose training of the machine learning model; the historical product data corresponding to a historical time duration; applying one or more calculations to the historical product data; combining the historical product data into a unified record; storing the unified record in a database; and inputting the aggregated product data into the machine learning model.
Lakhani further teaches determining relevant products to provide (i.e. [0003]), including the known technique of training of the machine learning model (Lakhani, see at least: “the demand estimator may be trained based on: determining a plurality of machine learning models including at least: a linear regression model, an elastic net model, an extreme gradient boosting (XGBoost) model [i.e. training of the machine learning model], and a random forecast (RF) model” [0079]) comprises:
the known techniques of generating aggregated product data based on the historical product data, the historical product data corresponding to a historical time duration (Lakhani, see at least: “the model training and demand estimation can be based on store clusters to increase signal to noise ratio, e.g. by aggregating estimated in-store demands of all stores in one cluster for an online item to determine whether to bring the online item to the stores in the cluster. An item sold in a high traffic store will likely have higher sales than if sold in a low traffic store. The store clusters can serve as store traffic proxy to represent different traffic patterns of different stores, when aggregating estimated in-store demands for an online item [i.e. generating aggregated product data based on the historical product data]” [0077] and “the ranking model 394 may be used to rank a plurality of store clusters based on their respective levels of transaction quantities (of one item, an item category, or a store department) [i.e. generating aggregated product data based on the historical product data]. In some examples, the ranking model 394 may be used to rank a plurality of online items based on their respective estimated demands in a future time period. In some examples, the ranking model 394 may be used to rank a plurality of in-store items being sold in a physical store based on sale numbers of the in-store items in a previous time period [i.e. the historical product data corresponding to a historical time duration]” [0055]), by:
applying one or more calculations to the historical product data (Lakhani, see at least: “the training dataset may be generated based on: clustering the plurality of physical stores into a plurality of clusters based on a quantity of transactions in each of the plurality of physical stores during a previous time period; determining a rank for each of the plurality of clusters based on their respective levels of transaction quantities [i.e. applying one or more calculations to the historical product data]; and incorporating the ranks of the plurality of clusters into the training dataset. In some embodiments, the model training and demand estimation can be based on store clusters to increase signal to noise ratio, e.g. by aggregating estimated in-store demands of all stores in one cluster for an online item to determine whether to bring the online item to the stores in the cluster” [0077]);
combining the historical product data into a unified record (Lakhani, see at least: “the model training and demand estimation can be based on store clusters to increase signal to noise ratio, e.g. by aggregating estimated in-store demands of all stores in one cluster for an online item to determine whether to bring the online item to the stores in the cluster [i.e. combining the historical product data into a unified record]. An item sold in a high traffic store will likely have higher sales than if sold in a low traffic store. The store clusters can serve as store traffic proxy to represent different traffic patterns of different stores, when aggregating estimated in-store demands for an online item” [0077] and “The demand estimation computing device 102 may parse the store related data 302 and the online purchase data 304 to generate store data 330 and user transaction data 340, respectively. In this example, the store data 330 may include, for each store, one or more of: a store ID 332 of the store, a store location 333 of the store, a historical demand 334 indicating historical demands for each displayed item in the store [i.e. into a unified record]” [0052]);
storing the unified record in a database (Lakhani, see at least: “The demand estimation computing device 102 may parse the store related data 302 and the online purchase data 304 to generate store data 330 and user transaction data 340, respectively. In this example, the store data 330 may include, for each store, one or more of: a store ID 332 of the store, a store location 333 of the store, a historical demand 334 indicating historical demands for each displayed item in the store [i.e. storing the unified record in a database]” [0052] and Fig. 3 indicates that the historical demand 334 is stored in database 116 [i.e. storing the unified record in a database]); and
inputting the aggregated product data into the machine learning model (Lakhani, see at least: “top similar items are determined for the item based on text similarity of their item descriptions relative to the item description of item, such that in-store demands for the top similar items in this year can be used to estimate an in-store demand for the item in next year based on the trained machine learning model. In some embodiments, the in-store demand for the item is computed also based on a local item popularity, which is computed based on a weighted average of online transactions that are delivered to delivery zip codes within a predetermined radius from the store's zip code, with each respective weight being an inverse of a distance between the respective delivery zip code and the store's zip code. In some embodiments, the in-store demand for the item may be computed also based on: store traffic data of an item category including the item in the store; online price data for the item; online historical demand data for the item; shipping speed data for online orders of the item; in-store price data for each of the top similar items; and in-store historical demand data for each of the top similar items” [0035]). It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine Hendlin with Lakhani for the reasons identified above with respect to claim 1.

Regarding claim 3, the combination of Hendlin/Lakhani/Abuomar teach the system of claim 2. Hendlin further discloses:
Hendlin does not explicitly disclose optimizing the machine learning model; wherein optimizing comprises tuning one or more hyperparameters associated with the machine learning model by applying Bayesian optimization to the one or more hyperparameters.
Lakhani further teaches determining relevant products to provide (i.e. [0003]), including optimizing the machine learning model (Lakhani, see at least: “the demand estimator may be trained based on: determining a plurality of machine learning models including at least: a linear regression model, an elastic net model, an extreme gradient boosting (XGBoost) model, and a random forecast (RF) model. For each of the plurality of models, a hyperparameter space is determined to include a plurality of hyperparameters for the machine learning model based on the training dataset. A plurality of evaluations can be performed, where each evaluation searches for a plurality of combinations of hyperparameters in the hyperparameter space. Then at least one optimized combination of hyperparameters is determined based on a Bayesian optimization [i.e. optimizing the machine learning model] and a validation using root mean square error (RMSE)” [0079]);
wherein optimizing comprises tuning one or more hyperparameters associated with the machine learning model by applying Bayesian optimization to the one or more hyperparameters (Lakhani, see at least: “the demand estimator may be trained based on: determining a plurality of machine learning models including at least: a linear regression model, an elastic net model, an extreme gradient boosting (XGBoost) model, and a random forecast (RF) model. For each of the plurality of models, a hyperparameter space is determined to include a plurality of hyperparameters for the machine learning model based on the training dataset. A plurality of evaluations can be performed, where each evaluation searches for a plurality of combinations of hyperparameters in the hyperparameter space. Then at least one optimized combination of hyperparameters is determined based on a Bayesian optimization [i.e. wherein optimizing comprises tuning one or more hyperparameters associated with the machine learning model by applying Bayesian optimization to the one or more hyperparameters] and a validation using root mean square error (RMSE)” [0079]).
It would have been obvious to one of ordinary skill in the art to include in the system, as taught by Hendlin, optimizing the machine learning model; wherein optimizing comprises tuning one or more hyperparameters associated with the machine learning model by applying Bayesian optimization to the one or more hyperparameters, as taught by Lakhani, since the claimed invention is merely a combination of old elements, and in the combination each element merely would have performed the same function as it did separately, and one of ordinary skill in the art would have recognized that the results of the combination were predictable. It further would have been obvious to one of ordinary skill in the art at the time of filing to modify Hendlin, to include the teachings of Lakhani, in order to recommend an assortment of items based on estimated demands (Lakhani, [0021]).

Regarding claim 4, the combination of Hendlin/Lakhani/Abuomar teach the system of claim 2. Hendlin further discloses:
Hendlin does not explicitly disclose the machine learning model being configured to generate the predicted amount of interactions based on at least one of historical product data or new product data.
Lakhani further teaches determining relevant products to provide (i.e. [0003]), including the known technique of the machine learning model being configured to generate the predicted amount of interactions based on at least one of historical product data or new product data (Lakhani, see at least: “the demand estimator [i.e. generate the predicted amount of interactions] may be trained based on: determining a plurality of machine learning models including at least: a linear regression model, an elastic net model, an extreme gradient boosting (XGBoost) model [i.e. wherein the machine learning model is configured to], and a random forecast (RF) model. For each of the plurality of models, a hyperparameter space is determined to include a plurality of hyperparameters for the machine learning model based on the training dataset. A plurality of evaluations can be performed, where each evaluation searches for a plurality of combinations of hyperparameters in the hyperparameter space. Then at least one optimized combination of hyperparameters is determined based on a Bayesian optimization and a validation using root mean square error (RMSE)” [0079] and “The demand estimation model 398 may be used to compute an estimated in-store demand for an online item [i.e. to generate the predicted amount of interactions], when it is offered for sale in a physical store in a future time period, e.g. based on one or more of: a similarity-based demand score for the online item based on similarity scores of the top K similar items; store traffic data of an item category including the online item in the physical store; a local item popularity of the online item for the physical store; online price data for the online item; online historical demand data for the online item; [i.e. based on at least one of historical product data or new product data] shipping speed data for online orders of the online item; in-store price data for each of the top K similar items; and in-store historical demand data for each of the top K similar items” [0057]). It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine Hendlin with Lakhani for the reasons identified above with respect to claim 1.

Regarding claim 5, the combination of Hendlin/Lakhani/Abuomar teach the system of claim 2. Hendlin further discloses:
-updating the model by (Hendlin, see at least: “the digital merchant content system 100 can receive an update to the product feed 450 indicating that a first product is out of stock. Similarly, the digital merchant content system 100 can receive additional social networking system data 458 indicating that a second product is generating increased sales [i.e. based on the at least one of the product feed or the product feed click count]. In response, the digital merchant content system 100 (e.g., via the ranking algorithm 460) [i.e. updating the model by] can modify the product display layout 442 (e.g., to remove the first product that is out of stock and to add the second product that is generating increased sales)” [0154]):
-obtaining from the second system: the product feed; and a product feed click count (Hendlin, see at least: “the ranking algorithm 460 can also apply criteria directed to product performance. For example, in one or more embodiments, the digital merchant content system 100 analyzes the social networking system data 458 [i.e. obtaining from the second system:] to identify metrics regarding product performance. To illustrate, the digital merchant content system 100 can determine a number of clicks a product receives on the social networking system [i.e. a product feed; and the product feed click count] (e.g., number of advertisements selected for a particular product), a number of purchases of a product via the social networking system, a number of views of a product via a social networking system, or a time duration that users view a product via a social networking system” [0134] and “the social networking system 104 can also perform the step 122 of storing the product feed (e.g., storing product information received from the product feed). In particular, the step 122 can include storing product information corresponding to a product feed at the social networking system 104 so that the social networking system 104 can access product information. For example, in one or more embodiments, the step 122 comprises storing a product feed and associating a product feed with a corresponding merchant (e.g., associating the product feed with an account of the merchant) [i.e. obtaining from the second system: the product feed]” [0065] and “the ranking algorithm can apply criteria directed to product performance (e.g., product sales, product clicks, product views, product discussion from social networking system data [i.e. obtaining from the second system: the product feed; and a product feed click count] or a product feed)” [0132]);
-providing at least one of the product feed or the product feed click count to the model (Hendlin, see at least: “the ranking algorithm [i.e. to the model] can apply criteria directed to product performance (e.g., product sales, product clicks, product views, product discussion from social networking system data [i.e. providing at least one of the product feed or the product feed click count] or a product feed)” [0132]); and
-updating in the model based on the at least one of the product feed or the product feed click count (Hendlin, see at least: “the digital merchant content system 100 can receive an update to the product feed 450 indicating that a first product is out of stock. Similarly, the digital merchant content system 100 can receive additional social networking system data 458 indicating that a second product is generating increased sales [i.e. based on the at least one of the product feed or the product feed click count]. In response, the digital merchant content system 100 (e.g., via the ranking algorithm 460) [i.e. updating the model] can modify the product display layout 442 (e.g., to remove the first product that is out of stock and to add the second product that is generating increased sales)” [0154] and “the ranking algorithm 460 can also apply criteria directed to product performance. For example, in one or more embodiments, the digital merchant content system 100 analyzes the social networking system data 458 to identify metrics regarding product performance. To illustrate, the digital merchant content system 100 can determine a number of clicks a product receives on the social networking system [i.e. based on the at least one of the product feed or the product feed click count] (e.g., number of advertisements selected for a particular product), a number of purchases of a product via the social networking system, a number of views of a product via a social networking system, or a time duration that users view a product via a social networking system” [0134]).

Hendlin does not explicitly disclose updating the machine learning model; model being a machine learning model; and updating one or more weights in the machine learning model by training the machine learning model.
Lakhani further teaches determining relevant products to provide (i.e. [0003]), including the known techniques of updating the machine learning model (Lakhani, see at least: “the demand estimation computing device 102 generates and/or updates different models for estimating in-store demand based on online data [i.e. updating the machine learning model]” [0037]) by:
providing at least one of the product feed or the product feed click count to the machine learning model (Lakhani, see at least: “the machine learning model data 390 [i.e. to the machine learning model] may include a similarity model 392, a ranking model 394, a local popularity model 396, a demand estimation model 398, and an assortment optimization model 399” [0054] and “the training dataset may comprise [i.e. to the machine learning model] at least one of the following features for each training item i of the plurality of training items: in-store related data 531 (including in-store price data and in-store historical demand data) for each in-store item j of the top K similar in-store items; online related data 532 (including online price data, online historical demand data, shipping speed data, and purchase channels) for online orders of the training item i [i.e. providing at least one of the product feed or the product feed click count]; store traffic data 533 of an item category including the training item i in each of the plurality of physical stores; a local item popularity 534 of the training item i for each of the plurality of physical stores” [0070] and “The web server 104 may transmit user session data related to a customer's activity (e.g., interactions) on the website. For example, a customer may operate one of customer computing devices 110, 112, 114 to initiate a web browser that is directed to the website hosted by the web server 104. The customer may, via the web browser, view item advertisements for items displayed on the website, and may click on item advertisements, [i.e. providing at least one of the product feed or the product feed click count] for example. The website may capture these activities as user session data, and transmit the user session data to the demand estimation computing device 102 over the communication network 118” [0032]); and
updating one or more weights in the machine learning model by training the machine learning model (Lakhani, see at least: “the demand estimator may be trained based on: determining a plurality of machine learning models including at least: a linear regression model, an elastic net model, an extreme gradient boosting (XGBoost) model, and a random forecast (RF) model. For each of the plurality of models, a hyperparameter space is determined to include a plurality of hyperparameters for the machine learning model [i.e. in the machine learning model] based on the training dataset” [0079] “FIG. 5 illustrates a process 500 for training [i.e. by training the machine learning model] a model for estimating in-store demand based on online data” [0065] and “a similarity-based demand score can be computed for the training item [i.e. by training the machine learning model] based on the similarity scores of the top K similar in-store items. For example, the similarity-based demand score can be computed based on: computing, for each similar item of the top K similar in-store items, a demand score indicating a historical in-store demand of the similar item during a previous time period; computing, for each similar item of the top K similar in-store items, an associated weight based on the similarity score of the similar item; computing a weighted sum of the demand scores of the top K similar in-store items, with their respective associated weights; computing a sum of the associated weights of the top K similar in-store items; and computing the similarity-based demand score for the training item based on a ratio between the weighted sum and the sum [i.e. updating one or more weights in the machine learning model]” [0068] and “the demand estimation computing device 102 generates and/or updates different models for estimating in-store demand based on online data [i.e. updating one or more weights in the machine learning model]” [0037]). It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine Hendlin with Lakhani for the reasons identified above with respect to claim 1.

Regarding claim 6, the combination of Hendlin/Lakhani/Abuomar teach the system of claim 5. Hendlin further discloses:
-wherein the model is updated daily (Hendlin, see at least: “the digital merchant content system 100 can check for updates utilizing a certain schedule (e.g., once a minute, twice an hour, or once a day) [i.e. updated daily]” [0115] and “The digital merchant content system 100 can also determine metrics of product performance via the product feed 450. For example, the product feed 450 can include updating information regarding product sales, product popularity, or other metrics (e.g., clicks on other websites, online purchases, or survey data). The ranking algorithm 460 can generate a score based on the extent to which a particular product corresponds to increased sales, popularity, clicks, purchases, or other metrics from the product feed 450 [i.e. wherein the model is updated]” [0136] and “the digital merchant content system 100 can receive an update to the product feed 450 indicating that a first product is out of stock. Similarly, the digital merchant content system 100 can receive additional social networking system data 458 indicating that a second product is generating increased sales. In response, the digital merchant content system 100 (e.g., via the ranking algorithm 460) can modify the product display layout 442 [i.e. wherein the model is updated] (e.g., to remove the first product that is out of stock and to add the second product that is generating increased sales)” [0154])
Hendlin does not explicitly disclose the model being a machine learning model.
Lakhani further teaches determining relevant products to provide (i.e. [0003]), including the known technique of a machine learning model being updated (Lakhani, see at least: “the demand estimator may be trained based on: determining a plurality of machine learning models including at least: a linear regression model, an elastic net model, an extreme gradient boosting (XGBoost) model [i.e. a machine learning model], and a random forecast (RF) model” [0079] and “the demand estimation computing device 102 generates and/or updates different models for estimating in-store demand based on online data [i.e. a machine learning model being updated]” [0037]). It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine Hendlin with Lakhani for the reasons identified above with respect to claim 1.

Regarding claim 7, the combination of Hendlin/Lakhani/Abuomar teach the system of claim 1. Hendlin further discloses:
-wherein the historical product data include at least one of vendor information, score information, product review information, customer information, sales information, or interactions corresponding to the webpage (Hendlin, see at least: “the ranking algorithm can apply criteria directed to product performance (e.g., product sales, product clicks, product views, product discussion from social networking system data or a product feed) [i.e. include at least one of vendor information, score information, product review information, customer information, sales information, or interactions corresponding to the webpage] … product information (e.g., a product order on an existing website, product emphasis, sale status, stock status from a product feed)” [0132] and “the ranking algorithm can utilize criteria directed to location, user history (e.g., a user's previous views, purchases, or clicks) [i.e. wherein the historical product data include at least one of vendor information, score information, product review information, customer information, sales information, or interactions corresponding to the webpage]” [0138]).

Regarding claim 8, the combination of Hendlin/Lakhani/Abuomar teach the system of claim 1. Hendlin further discloses:
-wherein a total amount of products in the second product set is no more than the capacity constraint (Hendlin, see at least: “the ranking algorithm 460 can rank the products 454a-454n by score and then select a number of products corresponding to a number of unpopulated elements in an unpopulated product display layout (i.e., the highest ranked products needed to fill the unpopulated elements) [i.e. wherein a total amount of products in the second product set is no more than the capacity constraint]” [0147] and Examiner notes that the number of unpopulated elements is the capacity constraint and the number of selected products doesn’t exceed the number of unpopulated elements [i.e. wherein a total amount of products in the second product set is no more than the capacity constraint]).
Hendlin in view of Lakhani does not explicitly teach a memory consumption corresponding to the total amount of products is no more than the capacity constraint.
Abuomar, however, teaches rules associated with a capacity (i.e. Col. 10 Ln. 9-16), including the known technique of a memory consumption corresponding to the products is no more than the capacity constraint (Abuomar, see at least: “FIG. 3 depicts an example rule 210 that reads: “If (Size(DataObject)>500 MB OR SalesRank(DataObject)>50) {Store in DataStore1} else {Store in DataStore2}”. The example rule 210 includes three expressions 214 and two threshold values 216. The first expression 214 describes a relationship between a data object characteristic 204 (the size of the data object 204) and a threshold value 216 (500 MB) [i.e. a memory consumption corresponding to the total amount of products is no more than the capacity constraint]” Col. 10 Ln. 9-16 and “a data object 104 may include information relating to an upcoming product that has not yet been released. Confidential portions of the data object 104 may include information relating to the product's identity, capabilities, parts, material specifications, images of the product, and so forth. Non-confidential portions of the data object 104 may include information relating to the product's price, quantity in stock, category, and so forth” Col. 11 Ln. 7-14). It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine Hendlin in view of Lakhani with Abuomar for the reasons identified above with respect to claim 1.

Regarding claim 9, the combination of Hendlin/Lakhani/Abuomar teach the system of claim 1. Hendlin further discloses:
-associating each product in the first set of products with a product feed key, the product feed key comprising a product identifier and a price (Hendlin, see at least: “the term “product feed” refers to a digital item defining product information. In particular, the term “product feed” includes an updating digital item that defines product information corresponding to products or services of a merchant over time. A product feed can take a variety of forms, for example, a product feed can comprise a database, spreadsheet, text file, or other electronic file [i.e. associating each product in the first set of products with a product feed key]. For instance, a merchant can maintain an updating database of products or services and provide the updating database as a product feed. In addition, the product feed can comprise a URL to a website. For example, a merchant can maintain an updating website that describes products or services and provide the website as a product feed” [0045] and “the digital merchant content system obtains a product feed comprising an updating database [i.e. the product feed key] with product information from a merchant (e.g., an updating database of product names [i.e. comprising a product identifier], product styles, product groups, product sales, product prices [i.e. comprising a price], and/or other product attributes)” [0033]); and
-mapping the product feed key to a product in the second system (Hendlin, see at least: “the merchant, via the merchant device 102, can provide the social networking system 104 with an indication of a digital location of an updating database that outlines product information corresponding to the merchant's products. The social networking system 104 can access the product feed by obtaining the updating database from the location identified by the merchant [i.e. mapping the product feed key to a product in the second system]” [0061] and “the social networking system 104 can also perform the step 122 of storing the product feed (e.g., storing product information received from the product feed). In particular, the step 122 can include storing product information corresponding to a product feed at the social networking system 104 so that the social networking system 104 can access product information. For example, in one or more embodiments, the step 122 comprises storing a product feed and associating a product feed with a corresponding merchant (e.g., associating the product feed with an account of the merchant) [i.e. mapping the product feed key to a product in the second system]” [0065]).

Regarding claim 10, the combination of Hendlin/Lakhani/Abuomar teach the system of claim 1. Hendlin further discloses:
Hendlin does not explicitly disclose the generated predicted amount of interactions corresponding to a future time period.
Lakhani further teaches determining relevant products to provide (i.e. [0003]), including the known technique of the generated predicted amount of interactions corresponding to a future time period (Lakhani, see at least: “the demand estimation computing device 102 may execute one or more models (e.g., programs or algorithms), such as a machine learning model, deep learning model, statistical model, etc., to compute estimated in-store demands for items [i.e. wherein the generated predicted amount of interactions], which are currently sold online but not in-store, when these items are offered for sale in the store 109 during a future time period, e.g. next month or next year [i.e. corresponds to a future time period]” [0034]). It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine Hendlin with Lakhani for the reasons identified above with respect to claim 1.

Regarding claim 11, Hendlin discloses a method for optimizing a product feed based on a model utilizing product data, comprising:
-obtaining historical product data corresponding to a first product set, the first product set comprising one or more products for display on a webpage associated with a first system (Hendlin, see at least: “the step 120 can also include obtaining a product feed from a website that includes a merchant's products or services [i.e. the first product set comprising one or more products for display on a webpage associated with a first system]. In particular, the social networking system 104 can obtain a product feed from a website by parsing the website and identifying product information. For example, in one or more embodiments, the social networking system 104 parses a website and identifies product names, product images and/or video, product prices, product sale status (e.g., products on sale), or product stock status” [0062] and “the ranking algorithm can apply criteria directed to product performance (e.g., product sales, product clicks, product views, product discussion from social networking system data or a product feed) [i.e. obtaining historical product data corresponding to a first product set] … product information (e.g., a product order on an existing website, product emphasis, sale status, stock status from a product feed)” [0132] and “the product feed 450 can include updating information regarding product sales, product popularity, or other metrics (e.g., clicks on other websites, online purchases, or survey data). The ranking algorithm 460 can generate a score based on the extent to which a particular product corresponds to increased sales, popularity, clicks, purchases, or other metrics from the product feed 450” [0136] “the ranking algorithm can utilize criteria directed to location, user history (e.g., a user's previous views, purchases, or clicks) [i.e. obtaining historical product data corresponding to a first product set]” [0138]);
-associating each product in the first set of products with a product feed key, the product feed key comprising a product identifier and a price (Hendlin, see at least: “the term “product feed” refers to a digital item defining product information. In particular, the term “product feed” includes an updating digital item that defines product information corresponding to products or services of a merchant over time. A product feed can take a variety of forms, for example, a product feed can comprise a database, spreadsheet, text file, or other electronic file [i.e. associating each product in the first set of products with a product feed key]. For instance, a merchant can maintain an updating database of products or services and provide the updating database as a product feed. In addition, the product feed can comprise a URL to a website. For example, a merchant can maintain an updating website that describes products or services and provide the website as a product feed” [0045] and “the digital merchant content system obtains a product feed comprising an updating database [i.e. the product feed key] with product information from a merchant (e.g., an updating database of product names [i.e. comprising a product identifier], product styles, product groups, product sales, product prices [i.e. comprising a price], and/or other product attributes)” [0033]);
-determining a capacity constraint associated with a second system configured to display information associated with the one or more products (Hendlin, see at least: “the ranking algorithm 460 can rank the products 454a-454n by score and then select a number of products corresponding to a number of unpopulated elements in an unpopulated product display layout (i.e., the highest ranked products needed to fill the unpopulated elements) [i.e. determining a capacity constraint associated with a second system configured to display information associated with the one or more products]” [0147] and “the social networking system 104 can also perform the step 138 of generating custom merchant content interfaces [i.e. associated with a second system configured to display information associated with the one or more products]. In particular, the social networking system 104 can generate a custom merchant content interface based on a custom merchant content template, a product feed, and/or social networking system data” [0071] Examiner notes that the number of unpopulated elements is the capacity constraint);
-generating an amount of interactions corresponding to at least one product in the first product set, wherein the generating is based on the historical product data (Hendlin, see at least: “The digital merchant content system 100 can also determine metrics of product performance via the product feed 450. For example, the product feed 450 can include updating information regarding product sales, product popularity, or other metrics (e.g., clicks on other websites, online purchases, or survey data). The ranking algorithm 460 can generate a score based on the extent to which a particular product corresponds to increased sales, popularity, clicks, purchases, or other metrics from the product feed 450 [i.e. generating an amount of interactions corresponding to at least one product in the first product set]” [0136] and “the ranking algorithm can apply criteria directed to product performance (e.g., product sales, product clicks, product views, product discussion from social networking system data or a product feed) [i.e. wherein the generating is based on the historical product data], user characteristics (e.g., demographic information, location information, or user history), product information (e.g., a product order on an existing website, product emphasis, sale status, stock status from a product feed), or a combination of such criteria” [0132]);
-selecting, based on the amount of interactions, a second product set, wherein the second product set includes a subset of the products from the first product set based on the capacity constraint (Hendlin, see at least: “the digital merchant content system 100 provides the product feed 450, the merchant preferences 456, and the social networking system data 458 to a ranking algorithm 460 and the ranking algorithm 460 generates selected products 462. [i.e. selecting a second product set, wherein the second product set includes a subset of the products from the first product] In particular, the ranking algorithm 460 generates the selected products 462 to utilize in populating unpopulated elements in the custom merchant content template 420 [i.e. based on the capacity constraint]. In one or more embodiments, the ranking algorithm 460 calculates a score corresponding to each product 454a-454n. In particular, the ranking algorithm 460 applies criteria to each of the products 454a-454n and calculates a score corresponding to each of the products 454a-454n based on whether (and/or to what extent) each of the products 454a-454n satisfy the criteria” [0131] and “the ranking algorithm can apply criteria directed to product performance (e.g., product sales, product clicks, product views, product discussion from social networking system data or a product feed) [i.e. based on the amount of interactions], user characteristics (e.g., demographic information, location information, or user history), product information (e.g., a product order on an existing website, product emphasis, sale status, stock status from a product feed), or a combination of such criteria” [0132] and “The digital merchant content system 100 can also determine metrics of product performance via the product feed 450. For example, the product feed 450 can include updating information regarding product sales, product popularity, or other metrics (e.g., clicks on other websites, online purchases, or survey data). The ranking algorithm 460 can generate a score based on the extent to which a particular product corresponds to increased sales, popularity, clicks, purchases, or other metrics from the product feed 450 [i.e. based on the amount of interactions]” [0136] Examiner notes that the number of unpopulated elements is the capacity constraint);
-mapping each product feed key of the second set of products to a product in the second system (Hendlin, see at least: “the merchant, via the merchant device 102, can provide the social networking system 104 with an indication of a digital location of an updating database that outlines product information corresponding to the merchant's products. The social networking system 104 can access the product feed by obtaining the updating database from the location identified by the merchant [i.e. mapping each product feed key of the second set of products to a product in the second system]” [0061] and “the social networking system 104 can also perform the step 122 of storing the product feed (e.g., storing product information received from the product feed) [i.e. to a product in the second system]. In particular, the step 122 can include storing product information corresponding to a product feed at the social networking system 104 so that the social networking system 104 can access product information. For example, in one or more embodiments, the step 122 comprises storing a product feed and associating a product feed with a corresponding merchant (e.g., associating the product feed with an account of the merchant) [i.e. mapping each product feed key of the second set of products to a product in the second system]” [0065] and “the digital merchant content system can also identify a product feed. For instance, in one or more embodiments, the digital merchant content system obtains a product feed comprising an updating database with product information from a merchant (e.g., an updating database of product names, product styles, product groups, product sales, product prices, and/or other product attributes)” [0033]); and
-presenting the second product set to the second system to update the product feed and display the updated product feed (Hendlin, see at least: “upon generating a score for the products 454a-454n, the ranking algorithm 460 can identify the selected products 462. In particular, in one or more embodiments, the ranking algorithm 460 identifies the selected products 462 by comparing scores corresponding to the products 454-454n. For example, the ranking algorithm 460 can rank the products 454a-454n by score and then select a number of products corresponding to a number of unpopulated elements in an unpopulated product display layout (i.e., the highest ranked products needed to fill the unpopulated elements) [i.e. presenting the second product set]” [0147] and “a digital merchant content system that provides one or more custom merchant content interfaces within a system and/or application that is separate from a merchant (e.g., a social networking system and/or application) [i.e. to the second system]. In particular, in one or more embodiments, the digital merchant content system generates custom merchant content interfaces that provide targeted, up-to-date, merchant content within a social networking system [i.e. to update the product feed and display the updated product feed]” [0026]).

Hendlin does not explicitly disclose the model being the model being a machine learning model trained on product data; generating, with a machine learning model, a predicted amount of interactions corresponding to at least one product, wherein the machine learning model is trained on the historical product data at training intervals based on the product feed; and the amount of interactions being a predicted amount of interactions.
Lakhani, however, teaches determining relevant products to provide (i.e. [0003]), including the known technique of a machine learning model trained on product data (Lakhani, see at least: “a machine learning model is trained based on data of shared items that are being sold both online and in-store by a retailer [i.e. a machine learning model trained on product data]. For a physical store of the retailer, inference items are determined from items being sold online but not in-store. An estimated demand is computed for each inference item to be offered for sale in the physical store in a future time period, based on the trained machine learning model and online data of the inference item. Based on the estimated demands for the inference items, recommended assortment data is generated for the physical store in the future time period” [0025]);
the known technique of generating, with a machine learning model, a predicted amount of interactions corresponding to at least one product, wherein the machine learning model is trained on the historical product data at training intervals based on the product feed (Lakhani, see at least: “the demand estimation computing device 102 may execute one or more models (e.g., programs or algorithms), such as a machine learning model, deep learning model, statistical model, etc., to compute estimated in-store demands for items, which are currently sold online but not in-store, when these items are offered for sale in the store 109 during a future time period, e.g. next month or next year [i.e. generating a predicted amount of interactions corresponding to at least one product]” [0034] and “the demand estimation computing device 102 may compute an estimated in-store demand for an item based on a machine learning model that is trained based on shared items sold both online and in-store … the in-store demand for the item may be computed also based on: store traffic data of an item category including the item in the store; online price data for the item; online historical demand data for the item [i.e. with a machine learning model, wherein the machine learning model is trained on the historical product data]” [0035] and “the demand estimation computing device 102 generates and/or updates different models for estimating in-store demand based on online data [i.e. at training intervals based on the product feed]. The models, when executed by the demand estimation computing device 102, allow the demand estimation computing device 102 to determine in-store demands and generate recommended assortment data for each store to refresh assortment for a future time period” [0037] and “the models may be trained following the process 500 in FIG. 5 every quarter or every year; and the assortment refresh may be automatically and periodically performed following the process 800 in FIG. 8, every month, every quarter or every year [i.e. at training intervals based on the product feed]” [0101]); and
the known technique of a predicted amount of interactions (Lakhani, see at least: “the demand estimation computing device 102 may execute one or more models (e.g., programs or algorithms), such as a machine learning model, deep learning model, statistical model, etc., to compute estimated in-store demands for items, which are currently sold online but not in-store, when these items are offered for sale in the store 109 during a future time period, e.g. next month or next year [i.e. a predicted amount of interactions]” [0034]). These known techniques are applicable to the method of Hendlin as they both share characteristics and capabilities, namely, they are directed to determining relevant products to provide.
It would have been recognized that applying the known techniques of a machine learning model trained on product data; generating, with a machine learning model, a predicted amount of interactions corresponding to at least one product, wherein the machine learning model is trained on the historical product data at training intervals based on the product feed; and a predicted amount of interactions, as taught by Lakhani, to the teachings of Hendlin would have yielded predictable results because the level of ordinary skill in the art demonstrated by the references applied shows the ability to incorporate such references into similar methods. Further, adding the modifications of a machine learning model trained on product data; generating, with a machine learning model, a predicted amount of interactions corresponding to at least one product, wherein the machine learning model is trained on the historical product data at training intervals based on the product feed; and a predicted amount of interactions, as taught by Lakhani, into the method of Hendlin would have been recognized by those of ordinary skill in the art as resulting in an improved method that would recommend an assortment of items based on estimated demands (Lakhani, [0021]).

Hendlin in view of Lakhani does not explicitly teach providing the capacity constraint to the first system, wherein the capacity constraint is based on a memory constraint of the second system.
Abuomar, however, teaches rules associated with a capacity (i.e. Col. 10 Ln. 9-16), including the known technique of providing the capacity constraint to the first system, wherein the capacity constraint is based on a memory constraint of the second system (Abuomar, see at least: “the rule data 122 may not include rules associated with a particular entity identifier 110 … The other rules may output to the entity 108 to solicit user input confirming or rejecting the rules [i.e. providing the capacity constraint to the first system]” Col. 6 Ln. 53-59 and “FIG. 3 depicts a scenario 300 illustrating a method for determining one or more rules 210 associated with a data object 104 and providing the data object 104 to a data store 102 indicated by the outcome of the rule(s) 210. At 302, a data object 104 may be received by a persistence server 106. In some implementations, the data object 104 may be provided by an entity 108, such as a service, a user, or another type of process or device. In other implementations, the data object 104 may be accessed by the persistence server 106. For example, the data object 104 may be stored in association with the persistence server 106 or in association with a data store 102 or computing device in communication with the persistence server 106. The data object 104 may include data object characteristics 204. For example, the data object characteristics 204 may include physical characteristics of the data object 104, such as a size of the data object (e.g., 2.3 MB) [i.e. wherein the capacity constraint is based on a memory constraint of the second system]” Col. 9 Ln. 15-31 and “FIG. 3 depicts an example rule 210 that reads: “If (Size(DataObject)>500 MB OR SalesRank(DataObject)>50) {Store in DataStore1} else {Store in DataStore2}”. The example rule 210 includes three expressions 214 and two threshold values 216. The first expression 214 describes a relationship between a data object characteristic 204 (the size of the data object 204) and a threshold value 216 (500 MB) [i.e. wherein the capacity constraint is based on a memory constraint of the second system]” Col. 10 Ln. 9-16 and “Each data object 104 may be stored in a particular data store 102 based on the characteristics of the data object 104 and on the data store characteristics 114 associated with the data stores 102. Data store characteristics 114 may include configurations associated with an associated data store 102, such as the access speed associated with stored data objects 104, a financial cost associated with using or accessing the data store 102, a computational cost associated with use or access of the data store 102, security features associated with the data store 102, storage capacity of the data store 102 [i.e. of the second system], restrictions regarding sizes or types of data objects 104 that may be stored in the data store 102, and so forth” Col. 5 Ln. 42-53 Examiner notes that entities 108 are the first system and the persistence server 106 is the second system). This known technique is applicable to the method of Hendlin in view of Lakhani as they both share characteristics and capabilities, namely, they are directed to rules associated with a capacity.
It would have been recognized that applying the known techniques of providing the capacity constraint to the first system, wherein the capacity constraint is based on a memory constraint of the second system, as taught by Abuomar, to the teachings of Hendlin in view of Lakhani would have yielded predictable results because the level of ordinary skill in the art demonstrated by the references applied shows the ability to incorporate such references into similar methods. Further, adding the modification of providing the capacity constraint to the first system, wherein the capacity constraint is based on a memory constraint of the second system method, as taught by Abuomar, into the method of Hendlin in view of Lakhani would have been recognized by those of ordinary skill in the art as resulting in an improved method that would determine rules associated with the received data objects (Abuomar, Col. 6 Ln. 23-26).

Regarding claim 12, the combination of Hendlin/Lakhani/Abuomar teach the method of claim 11. Hendlin further discloses:
-wherein updating of the model (Hendlin, see at least: “the digital merchant content system 100 can check for updates utilizing a certain schedule (e.g., once a minute, twice an hour, or once a day) [i.e. updating on a daily basis]” [0115] and “The digital merchant content system 100 can also determine metrics of product performance via the product feed 450. For example, the product feed 450 can include updating information regarding product sales, product popularity, or other metrics (e.g., clicks on other websites, online purchases, or survey data). The ranking algorithm 460 can generate a score based on the extent to which a particular product corresponds to increased sales, popularity, clicks, purchases, or other metrics from the product feed 450 [i.e. updating the model]” [0136] and “the digital merchant content system 100 can receive an update to the product feed 450 indicating that a first product is out of stock. Similarly, the digital merchant content system 100 can receive additional social networking system data 458 indicating that a second product is generating increased sales. In response, the digital merchant content system 100 (e.g., via the ranking algorithm 460) can modify the product display layout 442 [i.e. updating the model] (e.g., to remove the first product that is out of stock and to add the second product that is generating increased sales)” [0154]) comprises:
-generating, on a daily basis, aggregated product data based on the historical product data (Hendlin, see at least: “the digital merchant content system 100 can check for updates utilizing a certain schedule (e.g., once a minute, twice an hour, or once a day) [i.e. generating, on a daily basis,]” [0115] and “The digital merchant content system 100 can also determine metrics of product performance via the product feed 450. For example, the product feed 450 can include updating information regarding product sales, product popularity, or other metrics (e.g., clicks on other websites, online purchases, or survey data). The ranking algorithm 460 can generate a score based on the extent to which a particular product corresponds to increased sales, popularity, clicks, purchases, or other metrics from the product feed 450 [i.e. aggregated product data based on the historical product data]” [0136]), by:
-inputting the aggregated product data into the model daily (Hendlin, see at least: “The digital merchant content system 100 can also determine metrics of product performance via the product feed 450. For example, the product feed 450 can include updating information regarding product sales, product popularity, or other metrics (e.g., clicks on other websites, online purchases, or survey data). The ranking algorithm 460 can generate a score based on the extent to which a particular product corresponds to increased sales, popularity, clicks, purchases, or other metrics from the product feed 450 [i.e. inputting the aggregated product data into the model]” [0136] and “the database 300 can include media content items corresponding to products (e.g., a location of product image files or product video files), product specifications, product descriptions, product locations, product shipping items, product timing information (e.g., information indicating a time corresponding to seasonal products), product sale information (e.g., sale volume or sale rank compared to other products) [i.e. the aggregated product data]” [0114] and “the digital merchant content system 100 can check for updates utilizing a certain schedule (e.g., once a minute, twice an hour, or once a day) [i.e. daily]” [0115]).

Hendlin does not explicitly disclose training of the machine learning model; the historical product data corresponding to a historical time duration; applying one or more calculations to the historical product data; combining the historical product data into a unified record; storing the unified record in a database; and inputting the aggregated product data into the machine learning model.
Lakhani further teaches determining relevant products to provide (i.e. [0003]), including the known technique of training of the machine learning model (Lakhani, see at least: “the demand estimator may be trained based on: determining a plurality of machine learning models including at least: a linear regression model, an elastic net model, an extreme gradient boosting (XGBoost) model [i.e. training of the machine learning model], and a random forecast (RF) model” [0079]) comprises:
the known techniques of generating aggregated product data based on the historical product data, the historical product data corresponding to a historical time duration (Lakhani, see at least: “the model training and demand estimation can be based on store clusters to increase signal to noise ratio, e.g. by aggregating estimated in-store demands of all stores in one cluster for an online item to determine whether to bring the online item to the stores in the cluster. An item sold in a high traffic store will likely have higher sales than if sold in a low traffic store. The store clusters can serve as store traffic proxy to represent different traffic patterns of different stores, when aggregating estimated in-store demands for an online item [i.e. generating aggregated product data based on the historical product data]” [0077] and “the ranking model 394 may be used to rank a plurality of store clusters based on their respective levels of transaction quantities (of one item, an item category, or a store department) [i.e. generating aggregated product data based on the historical product data]. In some examples, the ranking model 394 may be used to rank a plurality of online items based on their respective estimated demands in a future time period. In some examples, the ranking model 394 may be used to rank a plurality of in-store items being sold in a physical store based on sale numbers of the in-store items in a previous time period [i.e. the historical product data corresponding to a historical time duration]” [0055]), by:
applying one or more calculations to the historical product data (Lakhani, see at least: “the training dataset may be generated based on: clustering the plurality of physical stores into a plurality of clusters based on a quantity of transactions in each of the plurality of physical stores during a previous time period; determining a rank for each of the plurality of clusters based on their respective levels of transaction quantities [i.e. applying one or more calculations to the historical product data]; and incorporating the ranks of the plurality of clusters into the training dataset. In some embodiments, the model training and demand estimation can be based on store clusters to increase signal to noise ratio, e.g. by aggregating estimated in-store demands of all stores in one cluster for an online item to determine whether to bring the online item to the stores in the cluster” [0077]);
combining the historical product data into a unified record (Lakhani, see at least: “the model training and demand estimation can be based on store clusters to increase signal to noise ratio, e.g. by aggregating estimated in-store demands of all stores in one cluster for an online item to determine whether to bring the online item to the stores in the cluster [i.e. combining the historical product data into a unified record]. An item sold in a high traffic store will likely have higher sales than if sold in a low traffic store. The store clusters can serve as store traffic proxy to represent different traffic patterns of different stores, when aggregating estimated in-store demands for an online item” [0077] and “The demand estimation computing device 102 may parse the store related data 302 and the online purchase data 304 to generate store data 330 and user transaction data 340, respectively. In this example, the store data 330 may include, for each store, one or more of: a store ID 332 of the store, a store location 333 of the store, a historical demand 334 indicating historical demands for each displayed item in the store [i.e. into a unified record]” [0052]);
storing the unified record in a database (Lakhani, see at least: “The demand estimation computing device 102 may parse the store related data 302 and the online purchase data 304 to generate store data 330 and user transaction data 340, respectively. In this example, the store data 330 may include, for each store, one or more of: a store ID 332 of the store, a store location 333 of the store, a historical demand 334 indicating historical demands for each displayed item in the store [i.e. storing the unified record in a database]” [0052] and Fig. 3 indicates that the historical demand 334 is stored in database 116 [i.e. storing the unified record in a database]); and
inputting the aggregated product data into the machine learning model (Lakhani, see at least: “top similar items are determined for the item based on text similarity of their item descriptions relative to the item description of item, such that in-store demands for the top similar items in this year can be used to estimate an in-store demand for the item in next year based on the trained machine learning model. In some embodiments, the in-store demand for the item is computed also based on a local item popularity, which is computed based on a weighted average of online transactions that are delivered to delivery zip codes within a predetermined radius from the store's zip code, with each respective weight being an inverse of a distance between the respective delivery zip code and the store's zip code. In some embodiments, the in-store demand for the item may be computed also based on: store traffic data of an item category including the item in the store; online price data for the item; online historical demand data for the item; shipping speed data for online orders of the item; in-store price data for each of the top similar items; and in-store historical demand data for each of the top similar items” [0035]). It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine Hendlin with Lakhani for the reasons identified above with respect to claim 11.

Regarding claim 13, the combination of Hendlin/Lakhani/Abuomar teach the method of claim 12. Hendlin further discloses:
Hendlin does not explicitly disclose optimizing the machine learning model; wherein optimizing comprises tuning one or more hyperparameters associated with the machine learning model by applying Bayesian optimization to the one or more hyperparameters.
Lakhani further teaches determining relevant products to provide (i.e. [0003]), including optimizing the machine learning model (Lakhani, see at least: “the demand estimator may be trained based on: determining a plurality of machine learning models including at least: a linear regression model, an elastic net model, an extreme gradient boosting (XGBoost) model, and a random forecast (RF) model. For each of the plurality of models, a hyperparameter space is determined to include a plurality of hyperparameters for the machine learning model based on the training dataset. A plurality of evaluations can be performed, where each evaluation searches for a plurality of combinations of hyperparameters in the hyperparameter space. Then at least one optimized combination of hyperparameters is determined based on a Bayesian optimization [i.e. optimizing the machine learning model] and a validation using root mean square error (RMSE)” [0079]);
wherein optimizing comprises tuning one or more hyperparameters associated with the machine learning model by applying Bayesian optimization to the one or more hyperparameters (Lakhani, see at least: “the demand estimator may be trained based on: determining a plurality of machine learning models including at least: a linear regression model, an elastic net model, an extreme gradient boosting (XGBoost) model, and a random forecast (RF) model. For each of the plurality of models, a hyperparameter space is determined to include a plurality of hyperparameters for the machine learning model based on the training dataset. A plurality of evaluations can be performed, where each evaluation searches for a plurality of combinations of hyperparameters in the hyperparameter space. Then at least one optimized combination of hyperparameters is determined based on a Bayesian optimization [i.e. wherein optimizing comprises tuning one or more hyperparameters associated with the machine learning model by applying Bayesian optimization to the one or more hyperparameters] and a validation using root mean square error (RMSE)” [0079]).
It would have been obvious to one of ordinary skill in the art to include in the method, as taught by Hendlin, optimizing the machine learning model; wherein optimizing comprises tuning one or more hyperparameters associated with the machine learning model by applying Bayesian optimization to the one or more hyperparameters, as taught by Lakhani, since the claimed invention is merely a combination of old elements, and in the combination each element merely would have performed the same function as it did separately, and one of ordinary skill in the art would have recognized that the results of the combination were predictable. It further would have been obvious to one of ordinary skill in the art at the time of filing to modify Hendlin, to include the teachings of Lakhani, in order to recommend an assortment of items based on estimated demands (Lakhani, [0021]).

Regarding claim 14, the combination of Hendlin/Lakhani/Abuomar teach the method of claim 12. Hendlin further discloses:
Hendlin does not explicitly disclose the machine learning model being configured to generate the predicted amount of interactions based on at least one of historical product data or new product data.
Lakhani further teaches determining relevant products to provide (i.e. [0003]), including the known technique of the machine learning model being configured to generate the predicted amount of interactions based on at least one of historical product data or new product data (Lakhani, see at least: “the demand estimator [i.e. generate the predicted amount of interactions] may be trained based on: determining a plurality of machine learning models including at least: a linear regression model, an elastic net model, an extreme gradient boosting (XGBoost) model [i.e. wherein the machine learning model is configured to], and a random forecast (RF) model. For each of the plurality of models, a hyperparameter space is determined to include a plurality of hyperparameters for the machine learning model based on the training dataset. A plurality of evaluations can be performed, where each evaluation searches for a plurality of combinations of hyperparameters in the hyperparameter space. Then at least one optimized combination of hyperparameters is determined based on a Bayesian optimization and a validation using root mean square error (RMSE)” [0079] and “The demand estimation model 398 may be used to compute an estimated in-store demand for an online item [i.e. to generate the predicted amount of interactions], when it is offered for sale in a physical store in a future time period, e.g. based on one or more of: a similarity-based demand score for the online item based on similarity scores of the top K similar items; store traffic data of an item category including the online item in the physical store; a local item popularity of the online item for the physical store; online price data for the online item; online historical demand data for the online item; [i.e. based on at least one of historical product data or new product data] shipping speed data for online orders of the online item; in-store price data for each of the top K similar items; and in-store historical demand data for each of the top K similar items” [0057]). It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine Hendlin with Lakhani for the reasons identified above with respect to claim 11.

Regarding claim 15, the combination of Hendlin/Lakhani/Abuomar teach the method of claim 12. Hendlin further discloses:
-updating the model by (Hendlin, see at least: “the digital merchant content system 100 can receive an update to the product feed 450 indicating that a first product is out of stock. Similarly, the digital merchant content system 100 can receive additional social networking system data 458 indicating that a second product is generating increased sales [i.e. based on the at least one of the product feed or the product feed click count]. In response, the digital merchant content system 100 (e.g., via the ranking algorithm 460) [i.e. updating the model by] can modify the product display layout 442 (e.g., to remove the first product that is out of stock and to add the second product that is generating increased sales)” [0154]):
-obtaining from the second system: the product feed; and a product feed click count (Hendlin, see at least: “the ranking algorithm 460 can also apply criteria directed to product performance. For example, in one or more embodiments, the digital merchant content system 100 analyzes the social networking system data 458 [i.e. obtaining from the second system:] to identify metrics regarding product performance. To illustrate, the digital merchant content system 100 can determine a number of clicks a product receives on the social networking system [i.e. a product feed; and the product feed click count] (e.g., number of advertisements selected for a particular product), a number of purchases of a product via the social networking system, a number of views of a product via a social networking system, or a time duration that users view a product via a social networking system” [0134] and “the social networking system 104 can also perform the step 122 of storing the product feed (e.g., storing product information received from the product feed). In particular, the step 122 can include storing product information corresponding to a product feed at the social networking system 104 so that the social networking system 104 can access product information. For example, in one or more embodiments, the step 122 comprises storing a product feed and associating a product feed with a corresponding merchant (e.g., associating the product feed with an account of the merchant) [i.e. obtaining from the second system: the product feed]” [0065] and “the ranking algorithm can apply criteria directed to product performance (e.g., product sales, product clicks, product views, product discussion from social networking system data [i.e. obtaining from the second system: the product feed; and a product feed click count] or a product feed)” [0132]);
-providing at least one of the product feed or the product feed click count to the model (Hendlin, see at least: “the ranking algorithm [i.e. to the model] can apply criteria directed to product performance (e.g., product sales, product clicks, product views, product discussion from social networking system data [i.e. providing at least one of the product feed or the product feed click count] or a product feed)” [0132]); and
-updating in the model based on the at least one of the product feed or the product feed click count (Hendlin, see at least: “the digital merchant content system 100 can receive an update to the product feed 450 indicating that a first product is out of stock. Similarly, the digital merchant content system 100 can receive additional social networking system data 458 indicating that a second product is generating increased sales [i.e. based on the at least one of the product feed or the product feed click count]. In response, the digital merchant content system 100 (e.g., via the ranking algorithm 460) [i.e. updating the model] can modify the product display layout 442 (e.g., to remove the first product that is out of stock and to add the second product that is generating increased sales)” [0154] and “the ranking algorithm 460 can also apply criteria directed to product performance. For example, in one or more embodiments, the digital merchant content system 100 analyzes the social networking system data 458 to identify metrics regarding product performance. To illustrate, the digital merchant content system 100 can determine a number of clicks a product receives on the social networking system [i.e. based on the at least one of the product feed or the product feed click count] (e.g., number of advertisements selected for a particular product), a number of purchases of a product via the social networking system, a number of views of a product via a social networking system, or a time duration that users view a product via a social networking system” [0134]).

Hendlin does not explicitly disclose updating the machine learning model; model being a machine learning model; and updating one or more weights in the machine learning model by training the machine learning model.
Lakhani further teaches determining relevant products to provide (i.e. [0003]), including the known techniques of updating the machine learning model (Lakhani, see at least: “the demand estimation computing device 102 generates and/or updates different models for estimating in-store demand based on online data [i.e. updating the machine learning model]” [0037]) by:
providing at least one of the product feed or the product feed click count to the machine learning model (Lakhani, see at least: “the machine learning model data 390 [i.e. to the machine learning model] may include a similarity model 392, a ranking model 394, a local popularity model 396, a demand estimation model 398, and an assortment optimization model 399” [0054] and “the training dataset may comprise [i.e. to the machine learning model] at least one of the following features for each training item i of the plurality of training items: in-store related data 531 (including in-store price data and in-store historical demand data) for each in-store item j of the top K similar in-store items; online related data 532 (including online price data, online historical demand data, shipping speed data, and purchase channels) for online orders of the training item i [i.e. providing at least one of the product feed or the product feed click count]; store traffic data 533 of an item category including the training item i in each of the plurality of physical stores; a local item popularity 534 of the training item i for each of the plurality of physical stores” [0070] and “The web server 104 may transmit user session data related to a customer's activity (e.g., interactions) on the website. For example, a customer may operate one of customer computing devices 110, 112, 114 to initiate a web browser that is directed to the website hosted by the web server 104. The customer may, via the web browser, view item advertisements for items displayed on the website, and may click on item advertisements, [i.e. providing at least one of the product feed or the product feed click count] for example. The website may capture these activities as user session data, and transmit the user session data to the demand estimation computing device 102 over the communication network 118” [0032]); and
updating one or more weights in the machine learning model by training the machine learning model (Lakhani, see at least: “the demand estimator may be trained based on: determining a plurality of machine learning models including at least: a linear regression model, an elastic net model, an extreme gradient boosting (XGBoost) model, and a random forecast (RF) model. For each of the plurality of models, a hyperparameter space is determined to include a plurality of hyperparameters for the machine learning model [i.e. in the machine learning model] based on the training dataset” [0079] “FIG. 5 illustrates a process 500 for training [i.e. by training the machine learning model] a model for estimating in-store demand based on online data” [0065] and “a similarity-based demand score can be computed for the training item [i.e. by training the machine learning model] based on the similarity scores of the top K similar in-store items. For example, the similarity-based demand score can be computed based on: computing, for each similar item of the top K similar in-store items, a demand score indicating a historical in-store demand of the similar item during a previous time period; computing, for each similar item of the top K similar in-store items, an associated weight based on the similarity score of the similar item; computing a weighted sum of the demand scores of the top K similar in-store items, with their respective associated weights; computing a sum of the associated weights of the top K similar in-store items; and computing the similarity-based demand score for the training item based on a ratio between the weighted sum and the sum [i.e. updating one or more weights in the machine learning model]” [0068] and “the demand estimation computing device 102 generates and/or updates different models for estimating in-store demand based on online data [i.e. updating one or more weights in the machine learning model]” [0037]). It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine Hendlin with Lakhani for the reasons identified above with respect to claim 11.

Regarding claim 16, the combination of Hendlin/Lakhani/Abuomar teach the method of claim 15. Hendlin further discloses:
-wherein the model is updated daily (Hendlin, see at least: “the digital merchant content system 100 can check for updates utilizing a certain schedule (e.g., once a minute, twice an hour, or once a day) [i.e. updated daily]” [0115] and “The digital merchant content system 100 can also determine metrics of product performance via the product feed 450. For example, the product feed 450 can include updating information regarding product sales, product popularity, or other metrics (e.g., clicks on other websites, online purchases, or survey data). The ranking algorithm 460 can generate a score based on the extent to which a particular product corresponds to increased sales, popularity, clicks, purchases, or other metrics from the product feed 450 [i.e. wherein the model is updated]” [0136] and “the digital merchant content system 100 can receive an update to the product feed 450 indicating that a first product is out of stock. Similarly, the digital merchant content system 100 can receive additional social networking system data 458 indicating that a second product is generating increased sales. In response, the digital merchant content system 100 (e.g., via the ranking algorithm 460) can modify the product display layout 442 [i.e. wherein the model is updated] (e.g., to remove the first product that is out of stock and to add the second product that is generating increased sales)” [0154])
Hendlin does not explicitly disclose the model being a machine learning model.
Lakhani further teaches determining relevant products to provide (i.e. [0003]), including the known technique of a machine learning model being updated (Lakhani, see at least: “the demand estimator may be trained based on: determining a plurality of machine learning models including at least: a linear regression model, an elastic net model, an extreme gradient boosting (XGBoost) model [i.e. a machine learning model], and a random forecast (RF) model” [0079] and “the demand estimation computing device 102 generates and/or updates different models for estimating in-store demand based on online data [i.e. a machine learning model being updated]” [0037]). It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine Hendlin with Lakhani for the reasons identified above with respect to claim 11.

Regarding claim 17, the combination of Hendlin/Lakhani/Abuomar teach the method of claim 11. Hendlin further discloses:
-wherein the historical product data include at least one of vendor information, score information, product review information, customer information, sales information, or interactions corresponding to the webpage (Hendlin, see at least: “the ranking algorithm can apply criteria directed to product performance (e.g., product sales, product clicks, product views, product discussion from social networking system data or a product feed) [i.e. include at least one of vendor information, score information, product review information, customer information, sales information, or interactions corresponding to the webpage] … product information (e.g., a product order on an existing website, product emphasis, sale status, stock status from a product feed)” [0132] and “the ranking algorithm can utilize criteria directed to location, user history (e.g., a user's previous views, purchases, or clicks) [i.e. wherein the historical product data include at least one of vendor information, score information, product review information, customer information, sales information, or interactions corresponding to the webpage]” [0138]).

Regarding claim 18, the combination of Hendlin/Lakhani/Abuomar teach the method of claim 11. Hendlin further discloses:
-wherein a total amount of products in the second product set is no more than the capacity constraint (Hendlin, see at least: “the ranking algorithm 460 can rank the products 454a-454n by score and then select a number of products corresponding to a number of unpopulated elements in an unpopulated product display layout (i.e., the highest ranked products needed to fill the unpopulated elements) [i.e. wherein a total amount of products in the second product set is no more than the capacity constraint]” [0147] and Examiner notes that the number of unpopulated elements is the capacity constraint and the number of selected products doesn’t exceed the number of unpopulated elements [i.e. wherein a total amount of products in the second product set is no more than the capacity constraint]).
Hendlin in view of Lakhani does not explicitly teach a memory consumption corresponding to the total amount of products is no more than the capacity constraint.
Abuomar, however, teaches rules associated with a capacity (i.e. Col. 10 Ln. 9-16), including the known technique of a memory consumption corresponding to the products is no more than the capacity constraint (Abuomar, see at least: “FIG. 3 depicts an example rule 210 that reads: “If (Size(DataObject)>500 MB OR SalesRank(DataObject)>50) {Store in DataStore1} else {Store in DataStore2}”. The example rule 210 includes three expressions 214 and two threshold values 216. The first expression 214 describes a relationship between a data object characteristic 204 (the size of the data object 204) and a threshold value 216 (500 MB) [i.e. a memory consumption corresponding to the total amount of products is no more than the capacity constraint]” Col. 10 Ln. 9-16 and “a data object 104 may include information relating to an upcoming product that has not yet been released. Confidential portions of the data object 104 may include information relating to the product's identity, capabilities, parts, material specifications, images of the product, and so forth. Non-confidential portions of the data object 104 may include information relating to the product's price, quantity in stock, category, and so forth” Col. 11 Ln. 7-14). It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine Hendlin in view of Lakhani with Abuomar for the reasons identified above with respect to claim 11.

Regarding claim 19, the combination of Hendlin/Lakhani/Abuomar teach the method of claim 11. Hendlin further discloses:
-associating each product in the first set of products with a product feed key, the product feed key comprising a product identifier and a price (Hendlin, see at least: “the term “product feed” refers to a digital item defining product information. In particular, the term “product feed” includes an updating digital item that defines product information corresponding to products or services of a merchant over time. A product feed can take a variety of forms, for example, a product feed can comprise a database, spreadsheet, text file, or other electronic file [i.e. associating each product in the first set of products with a product feed key]. For instance, a merchant can maintain an updating database of products or services and provide the updating database as a product feed. In addition, the product feed can comprise a URL to a website. For example, a merchant can maintain an updating website that describes products or services and provide the website as a product feed” [0045] and “the digital merchant content system obtains a product feed comprising an updating database [i.e. the product feed key] with product information from a merchant (e.g., an updating database of product names [i.e. comprising a product identifier], product styles, product groups, product sales, product prices [i.e. comprising a price], and/or other product attributes)” [0033]); and
-mapping the product feed key to a product in the second system (Hendlin, see at least: “the merchant, via the merchant device 102, can provide the social networking system 104 with an indication of a digital location of an updating database that outlines product information corresponding to the merchant's products. The social networking system 104 can access the product feed by obtaining the updating database from the location identified by the merchant [i.e. mapping the product feed key to a product in the second system]” [0061] and “the social networking system 104 can also perform the step 122 of storing the product feed (e.g., storing product information received from the product feed). In particular, the step 122 can include storing product information corresponding to a product feed at the social networking system 104 so that the social networking system 104 can access product information. For example, in one or more embodiments, the step 122 comprises storing a product feed and associating a product feed with a corresponding merchant (e.g., associating the product feed with an account of the merchant) [i.e. mapping the product feed key to a product in the second system]” [0065]).

Regarding claim 20, Hendlin discloses a non-transitory computer readable medium including instructions that are executable by one or more processors to cause a system to perform a method (Hendlin, see at least: “The components 702-716 and their corresponding elements can comprise software, hardware, or both. For example, the components 702-716 and their corresponding elements can comprise one or more instructions stored on a computer-readable storage medium and executable by processors of one or more computing devices [i.e. instructions that are executable by one or more processors to cause a system to perform a method]” [0195]), the method comprising:
-obtaining historical product data corresponding to a first product set, the first product set comprising one or more products for display on a webpage associated with the system (Hendlin, see at least: “the step 120 can also include obtaining a product feed from a website that includes a merchant's products or services [i.e. wherein the first product set comprising one or more products for display on a webpage associated with the system]. In particular, the social networking system 104 can obtain a product feed from a website by parsing the website and identifying product information. For example, in one or more embodiments, the social networking system 104 parses a website and identifies product names, product images and/or video, product prices, product sale status (e.g., products on sale), or product stock status” [0062] and “the ranking algorithm can apply criteria directed to product performance (e.g., product sales, product clicks, product views, product discussion from social networking system data or a product feed) [i.e. obtaining historical product data corresponding to a first product set] … product information (e.g., a product order on an existing website, product emphasis, sale status, stock status from a product feed)” [0132] and “the product feed 450 can include updating information regarding product sales, product popularity, or other metrics (e.g., clicks on other websites, online purchases, or survey data). The ranking algorithm 460 can generate a score based on the extent to which a particular product corresponds to increased sales, popularity, clicks, purchases, or other metrics from the product feed 450” [0136] and “the ranking algorithm can utilize criteria directed to location, user history (e.g., a user's previous views, purchases, or clicks) [i.e. obtaining historical product data corresponding to a first product set]” [0138]);
-associating each product in the first set of products with a product feed key, the product feed key comprising a product identifier and a price (Hendlin, see at least: “the term “product feed” refers to a digital item defining product information. In particular, the term “product feed” includes an updating digital item that defines product information corresponding to products or services of a merchant over time. A product feed can take a variety of forms, for example, a product feed can comprise a database, spreadsheet, text file, or other electronic file [i.e. associating each product in the first set of products with a product feed key]. For instance, a merchant can maintain an updating database of products or services and provide the updating database as a product feed. In addition, the product feed can comprise a URL to a website. For example, a merchant can maintain an updating website that describes products or services and provide the website as a product feed” [0045] and “the digital merchant content system obtains a product feed comprising an updating database [i.e. the product feed key] with product information from a merchant (e.g., an updating database of product names [i.e. comprising a product identifier], product styles, product groups, product sales, product prices [i.e. comprising a price], and/or other product attributes)” [0033]);
-determining a capacity constraint associated with a second system configured to display information associated with the one or more products (Hendlin, see at least: “the ranking algorithm 460 can rank the products 454a-454n by score and then select a number of products corresponding to a number of unpopulated elements in an unpopulated product display layout (i.e., the highest ranked products needed to fill the unpopulated elements) [i.e. determining a capacity constraint associated with a second system configured to display information associated with the one or more products]” [0147] and “the social networking system 104 can also perform the step 138 of generating custom merchant content interfaces [i.e. associated with a second system configured to display information associated with the one or more products]. In particular, the social networking system 104 can generate a custom merchant content interface based on a custom merchant content template, a product feed, and/or social networking system data” [0071] Examiner notes that the number of unpopulated elements is the capacity constraint);
-wherein the capacity constraint comprises a numerical limit (Hendlin, see at least: “the ranking algorithm 460 can rank the products 454a-454n by score and then select a number of products corresponding to a number of unpopulated elements in an unpopulated product display layout (i.e., the highest ranked products needed to fill the unpopulated elements) [i.e. wherein the capacity constraint comprises a numerical limit]” [0147] Examiner notes that the number of unpopulated elements is the capacity constraint [i.e. wherein the capacity constraint comprises a numerical limit]);
-generating amount of interactions corresponding to at least one product in the first product set (Hendlin, see at least: “The digital merchant content system 100 can also determine metrics of product performance via the product feed 450. For example, the product feed 450 can include updating information regarding product sales, product popularity, or other metrics (e.g., clicks on other websites, online purchases, or survey data). The ranking algorithm 460 can generate a score based on the extent to which a particular product corresponds to increased sales, popularity, clicks, purchases, or other metrics from the product feed 450 [i.e. generating an amount of interactions corresponding to at least one product in the first product set]” [0136]);
-updating, at intervals based on the product feed, the model by (Hendlin, see at least: “the digital merchant content system 100 can check for updates utilizing a certain schedule (e.g., once a minute, twice an hour, or once a day) [i.e. updating, at intervals based on the product feed]” [0115] and “The digital merchant content system 100 can also determine metrics of product performance via the product feed 450. For example, the product feed 450 can include updating information regarding product sales, product popularity, or other metrics (e.g., clicks on other websites, online purchases, or survey data). The ranking algorithm 460 can generate a score based on the extent to which a particular product corresponds to increased sales, popularity, clicks, purchases, or other metrics from the product feed 450 [i.e. updating the model]” [0136] and “the digital merchant content system 100 can receive an update to the product feed 450 indicating that a first product is out of stock. Similarly, the digital merchant content system 100 can receive additional social networking system data 458 indicating that a second product is generating increased sales. In response, the digital merchant content system 100 (e.g., via the ranking algorithm 460) can modify the product display layout 442 [i.e. updating the model] (e.g., to remove the first product that is out of stock and to add the second product that is generating increased sales)” [0154]):
-generating aggregated product data based on the historical product data (Hendlin, see at least: “the database 300 can include media content items corresponding to products (e.g., a location of product image files or product video files), product specifications, product descriptions, product locations, product shipping items, product timing information (e.g., information indicating a time corresponding to seasonal products), product sale information (e.g., sale volume or sale rank compared to other products) [i.e. generating aggregated product data based on the historical product data]” [0114] and “The digital merchant content system 100 can also determine metrics of product performance via the product feed 450 [i.e. generating aggregated product data]. For example, the product feed 450 can include updating information regarding product sales, product popularity, or other metrics (e.g., clicks on other websites, online purchases, or survey data)” [0136] and “the ranking algorithm can utilize criteria directed to location, user history (e.g., a user's previous views, purchases, or clicks) [i.e. based on the historical product data]” [0138]);
-inputting the aggregated product data into the model (Hendlin, see at least: “The digital merchant content system 100 can also determine metrics of product performance via the product feed 450. For example, the product feed 450 can include updating information regarding product sales, product popularity, or other metrics (e.g., clicks on other websites, online purchases, or survey data). The ranking algorithm 460 can generate a score based on the extent to which a particular product corresponds to increased sales, popularity, clicks, purchases, or other metrics from the product feed 450 [i.e. inputting the aggregated product data into the model]” [0136] and “the database 300 can include media content items corresponding to products (e.g., a location of product image files or product video files), product specifications, product descriptions, product locations, product shipping items, product timing information (e.g., information indicating a time corresponding to seasonal products), product sale information (e.g., sale volume or sale rank compared to other products) [i.e. the aggregated product data]” [0114]);
-obtaining from the second system: the product feed; and a product feed click count (Hendlin, see at least: “the ranking algorithm 460 can also apply criteria directed to product performance. For example, in one or more embodiments, the digital merchant content system 100 analyzes the social networking system data 458 [i.e. obtaining from the second system:] to identify metrics regarding product performance. To illustrate, the digital merchant content system 100 can determine a number of clicks a product receives on the social networking system [i.e. a product feed; and a product feed click count] (e.g., number of advertisements selected for a particular product), a number of purchases of a product via the social networking system, a number of views of a product via a social networking system, or a time duration that users view a product via a social networking system” [0134] and “the social networking system 104 can also perform the step 122 of storing the product feed (e.g., storing product information received from the product feed). In particular, the step 122 can include storing product information corresponding to a product feed at the social networking system 104 so that the social networking system 104 can access product information. For example, in one or more embodiments, the step 122 comprises storing a product feed and associating a product feed with a corresponding merchant (e.g., associating the product feed with an account of the merchant) [i.e. obtaining from the second system: a product feed]” [0065] and “the ranking algorithm can apply criteria directed to product performance (e.g., product sales, product clicks, product views, product discussion from social networking system data [i.e. obtaining from the second system: a product feed; and a product feed click count] or a product feed)” [0132]);
-providing at least one of the product feed or the product feed click count to the model (Hendlin, see at least: “the ranking algorithm [i.e. to the model] can apply criteria directed to product performance (e.g., product sales, product clicks, product views, product discussion from social networking system data [i.e. providing at least one of the product feed or the product feed click count] or a product feed)” [0132]); and
-updating the model based on the at least one of the product feed or the product feed click count (Hendlin, see at least: “the digital merchant content system 100 can receive an update to the product feed 450 indicating that a first product is out of stock. Similarly, the digital merchant content system 100 can receive additional social networking system data 458 indicating that a second product is generating increased sales [i.e. based on the at least one of the product feed or the product feed click count]. In response, the digital merchant content system 100 (e.g., via the ranking algorithm 460) [i.e. updating the model] can modify the product display layout 442 (e.g., to remove the first product that is out of stock and to add the second product that is generating increased sales)” [0154] and “the ranking algorithm 460 can also apply criteria directed to product performance. For example, in one or more embodiments, the digital merchant content system 100 analyzes the social networking system data 458 to identify metrics regarding product performance. To illustrate, the digital merchant content system 100 can determine a number of clicks a product receives on the social networking system [i.e. based on the at least one of the product feed or the product feed click count] (e.g., number of advertisements selected for a particular product), a number of purchases of a product via the social networking system, a number of views of a product via a social networking system, or a time duration that users view a product via a social networking system” [0134]);
-selecting, based on the amount of interactions, a second product set, wherein the second product set includes a subset of the one or more products from the first product set based on the capacity constraint (Hendlin, see at least: “the digital merchant content system 100 provides the product feed 450, the merchant preferences 456, and the social networking system data 458 to a ranking algorithm 460 and the ranking algorithm 460 generates selected products 462. [i.e. selecting a second product set, wherein the second product set includes a subset of the products from the first product] In particular, the ranking algorithm 460 generates the selected products 462 to utilize in populating unpopulated elements in the custom merchant content template 420 [i.e. based on the capacity constraint]. In one or more embodiments, the ranking algorithm 460 calculates a score corresponding to each product 454a-454n. In particular, the ranking algorithm 460 applies criteria to each of the products 454a-454n and calculates a score corresponding to each of the products 454a-454n based on whether (and/or to what extent) each of the products 454a-454n satisfy the criteria” [0131] and “the ranking algorithm can apply criteria directed to product performance (e.g., product sales, product clicks, product views, product discussion from social networking system data or a product feed) [i.e. based on the amount of interactions], user characteristics (e.g., demographic information, location information, or user history), product information (e.g., a product order on an existing website, product emphasis, sale status, stock status from a product feed), or a combination of such criteria” [0132] and “The digital merchant content system 100 can also determine metrics of product performance via the product feed 450. For example, the product feed 450 can include updating information regarding product sales, product popularity, or other metrics (e.g., clicks on other websites, online purchases, or survey data). The ranking algorithm 460 can generate a score based on the extent to which a particular product corresponds to increased sales, popularity, clicks, purchases, or other metrics from the product feed 450 [i.e. based on the amount of interactions]” [0136] Examiner notes that the number of unpopulated elements is the capacity constraint);
-mapping each product feed key of the second set of products to a product in the second system (Hendlin, see at least: “the merchant, via the merchant device 102, can provide the social networking system 104 with an indication of a digital location of an updating database that outlines product information corresponding to the merchant's products. The social networking system 104 can access the product feed by obtaining the updating database from the location identified by the merchant [i.e. mapping each product feed key of the second set of products to a product in the second system]” [0061] and “the social networking system 104 can also perform the step 122 of storing the product feed (e.g., storing product information received from the product feed) [i.e. to a product in the second system]. In particular, the step 122 can include storing product information corresponding to a product feed at the social networking system 104 so that the social networking system 104 can access product information. For example, in one or more embodiments, the step 122 comprises storing a product feed and associating a product feed with a corresponding merchant (e.g., associating the product feed with an account of the merchant) [i.e. mapping each product feed key of the second set of products to a product in the second system]” [0065] and “the digital merchant content system can also identify a product feed. For instance, in one or more embodiments, the digital merchant content system obtains a product feed comprising an updating database with product information from a merchant (e.g., an updating database of product names, product styles, product groups, product sales, product prices, and/or other product attributes)” [0033]); and
-presenting the second product set to the second system for display (Hendlin, see at least: “upon generating a score for the products 454a-454n, the ranking algorithm 460 can identify the selected products 462. In particular, in one or more embodiments, the ranking algorithm 460 identifies the selected products 462 by comparing scores corresponding to the products 454-454n. For example, the ranking algorithm 460 can rank the products 454a-454n by score and then select a number of products corresponding to a number of unpopulated elements in an unpopulated product display layout (i.e., the highest ranked products needed to fill the unpopulated elements) [i.e. presenting the second product set]” [0147] and “a digital merchant content system that provides one or more custom merchant content interfaces within a system and/or application that is separate from a merchant (e.g., a social networking system and/or application) [i.e. to the second system]. In particular, in one or more embodiments, the digital merchant content system generates custom merchant content interfaces that provide targeted, up-to-date, merchant content within a social networking system” [0026]).

Hendlin does not explicitly disclose generating a gradient boosting machine regression model configured to generate a predicted amount of interactions; training at training intervals based on the product feed, the gradient boosting machine regression model; the model being the gradient boosting machine regression model; the model being a gradient boosting machine regression model; updating one or more weights in gradient boosting machine regression;
optimizing the trained gradient boosting machine regression model by applying Bayesian optimization to one or more hyperparameters associated with the trained gradient boosting machine regression model; generating, with the optimized gradient boosting machine regression model, the predicted amount of interactions; wherein the predicted amount of interactions corresponds to a future time interval; the amount of interactions being a predicted amount of interactions; and presenting the second product set being for display during the future time interval.
Lakhani, however, teaches determining relevant products to provide (i.e. [0003]), including the known technique of generating a gradient boosting machine regression model configured to generate a predicted amount of interactions (Lakhani, see at least: “the demand estimator may be trained based on: determining a plurality of machine learning models including at least: a linear regression model, an elastic net model, an extreme gradient boosting (XGBoost) model [i.e. generating a gradient boosting machine regression model], and a random forecast (RF) model” [0079] and “the demand estimation computing device 102 may execute one or more models (e.g., programs or algorithms), such as a machine learning model, deep learning model, statistical model, etc., to compute estimated in-store demands for items, which are currently sold online but not in-store, when these items are offered for sale in the store 109 during a future time period, e.g. next month or next year [i.e. configured to generate a predicted amount of interactions]” [0034]);
the known techniques of training at training intervals based on the product feed, the gradient boosting machine regression model (Lakhani, see at least: “the demand estimator may be trained based on: determining a plurality of machine learning models including at least: a linear regression model, an elastic net model, an extreme gradient boosting (XGBoost) model [i.e. training the gradient boosting machine regression model], and a random forecast (RF) model” [0079] and “the demand estimation computing device 102 generates and/or updates different models for estimating in-store demand based on online data [i.e. at training intervals based on the product feed]. The models, when executed by the demand estimation computing device 102, allow the demand estimation computing device 102 to determine in-store demands and generate recommended assortment data for each store to refresh assortment for a future time period” [0037] and “the models may be trained following the process 500 in FIG. 5 every quarter or every year; and the assortment refresh may be automatically and periodically performed following the process 800 in FIG. 8, every month, every quarter or every year [i.e. at training intervals based on the product feed]” [0101]) by:
inputting the aggregated product data into the gradient boosting machine regression model (Lakhani, see at least: “the training dataset may comprise at least one of the following features for each training item i of the plurality of training items: in-store related data 531 (including in-store price data and in-store historical demand data) for each in-store item j of the top K similar in-store items; online related data 532 (including online price data, online historical demand data, shipping speed data, and purchase channels) for online orders of the training item i [i.e. inputting the aggregated product data]; store traffic data 533 of an item category including the training item i in each of the plurality of physical stores; a local item popularity 534 of the training item i for each of the plurality of physical stores” [0070] and “the demand estimator may be trained based on: determining a plurality of machine learning models including at least: a linear regression model, an elastic net model, an extreme gradient boosting (XGBoost) model [i.e. into the gradient boosting machine regression model], and a random forecast (RF) model” [0079]);
providing at least one of the product feed or the product feed click count to the gradient boosting machine regression model (Lakhani, see at least: “the machine learning model data 390 [i.e. to the gradient boosting machine regression model] may include a similarity model 392, a ranking model 394, a local popularity model 396, a demand estimation model 398, and an assortment optimization model 399” [0054] and “the training dataset may comprise [i.e. to the gradient boosting machine regression model] at least one of the following features for each training item i of the plurality of training items: in-store related data 531 (including in-store price data and in-store historical demand data) for each in-store item j of the top K similar in-store items; online related data 532 (including online price data, online historical demand data, shipping speed data, and purchase channels) for online orders of the training item i [i.e. providing at least one of the product feed or the product feed click count]; store traffic data 533 of an item category including the training item i in each of the plurality of physical stores; a local item popularity 534 of the training item i for each of the plurality of physical stores” [0070] and “The web server 104 may transmit user session data related to a customer's activity (e.g., interactions) on the website. For example, a customer may operate one of customer computing devices 110, 112, 114 to initiate a web browser that is directed to the website hosted by the web server 104. The customer may, via the web browser, view item advertisements for items displayed on the website, and may click on item advertisements, [i.e. providing at least one of the product feed or the product feed click count] for example. The website may capture these activities as user session data, and transmit the user session data to the demand estimation computing device 102 over the communication network 118” [0032]);
updating one or more weights in the gradient boosting machine regression (Lakhani, see at least: “the demand estimator may be trained based on: determining a plurality of machine learning models including at least: a linear regression model, an elastic net model, an extreme gradient boosting (XGBoost) model, and a random forecast (RF) model. For each of the plurality of models, a hyperparameter space is determined to include a plurality of hyperparameters for the machine learning model [i.e. in the gradient boosting machine regression] based on the training dataset” [0079] “FIG. 5 illustrates a process 500 for training a model for estimating in-store demand based on online data” [0065] and “a similarity-based demand score can be computed for the training item based on the similarity scores of the top K similar in-store items. For example, the similarity-based demand score can be computed based on: computing, for each similar item of the top K similar in-store items, a demand score indicating a historical in-store demand of the similar item during a previous time period; computing, for each similar item of the top K similar in-store items, an associated weight based on the similarity score of the similar item; computing a weighted sum of the demand scores of the top K similar in-store items, with their respective associated weights; computing a sum of the associated weights of the top K similar in-store items; and computing the similarity-based demand score for the training item based on a ratio between the weighted sum and the sum [i.e. updating one or more weights in the gradient boosting machine regression]” [0068] and “the demand estimation computing device 102 generates and/or updates different models for estimating in-store demand based on online data [i.e. updating one or more weights in the gradient boosting machine regression]” [0037]); and
optimizing the trained gradient boosting machine regression model by applying Bayesian optimization to one or more hyperparameters associated with the trained gradient boosting machine regression model (Lakhani, see at least: “the demand estimator may be trained based on: determining a plurality of machine learning models including at least: a linear regression model, an elastic net model, an extreme gradient boosting (XGBoost) model, and a random forecast (RF) model. For each of the plurality of models, a hyperparameter space is determined to include a plurality of hyperparameters for the machine learning model based on the training dataset. A plurality of evaluations can be performed, where each evaluation searches for a plurality of combinations of hyperparameters in the hyperparameter space. Then at least one optimized combination of hyperparameters is determined based on a Bayesian optimization [i.e. optimizing the trained gradient boosting machine regression model by applying Bayesian optimization to one or more hyperparameters associated with the trained gradient boosting machine regression model] and a validation using root mean square error (RMSE)” [0079]);
the known technique of generating, with the optimized gradient boosting machine regression model, the predicted amount of interactions (Lakhani, see at least: “the demand estimator [i.e. generating the predicted amount of interactions] may be trained based on: determining a plurality of machine learning models including at least: a linear regression model, an elastic net model, an extreme gradient boosting (XGBoost) model [i.e. with the optimized gradient boosting machine regression model], and a random forecast (RF) model. For each of the plurality of models, a hyperparameter space is determined to include a plurality of hyperparameters for the machine learning model based on the training dataset. A plurality of evaluations can be performed, where each evaluation searches for a plurality of combinations of hyperparameters in the hyperparameter space. Then at least one optimized combination of hyperparameters is determined based on a Bayesian optimization and a validation using root mean square error (RMSE)” [0079]);
the known technique of the predicted amount of interactions corresponding to a future time interval (Lakhani, see at least: “the demand estimation computing device 102 may execute one or more models (e.g., programs or algorithms), such as a machine learning model, deep learning model, statistical model, etc., to compute estimated in-store demands for items, which are currently sold online but not in-store, when these items are offered for sale in the store 109 during a future time period, e.g. next month or next year [i.e. wherein the predicted amount of interactions corresponds to a future time interval]” [0034]);
the known technique of a predicted amount of interactions (Lakhani, see at least: “the demand estimation computing device 102 may execute one or more models (e.g., programs or algorithms), such as a machine learning model, deep learning model, statistical model, etc., to compute estimated in-store demands for items, which are currently sold online but not in-store, when these items are offered for sale in the store 109 during a future time period, e.g. next month or next year [i.e. a predicted amount of interactions]” [0034]); and
the known technique of presenting products for display during the future time interval (Lakhani, see at least: “the demand estimation computing device 102 may execute one or more models (e.g., programs or algorithms), such as a machine learning model, deep learning model, statistical model, etc., to compute estimated in-store demands for items, which are currently sold online but not in-store, when these items are offered for sale in the store 109 [i.e. presenting products for display] during a future time period, e.g. next month or next year [i.e. for display during the future time interval]” [0034]). These known techniques are applicable to the non-transitory computer readable medium of Hendlin as they both share characteristics and capabilities, namely, they are directed to determining relevant products to provide.
It would have been recognized that applying the known techniques of generating a gradient boosting machine regression model configured to generate a predicted amount of interactions; training at training intervals based on the product feed, the gradient boosting machine regression model by: inputting the aggregated product data into the gradient boosting machine regression model; providing at least one of the product feed or the product feed click count to the gradient boosting machine regression model; updating one or more weights in the gradient boosting machine regression; and optimizing the trained gradient boosting machine regression model by applying Bayesian optimization to one or more hyperparameters associated with the trained gradient boosting machine regression model; generating, with the optimized gradient boosting machine regression model, the predicted amount of interactions; the predicted amount of interactions corresponding to a future time interval; a predicted amount of interactions; and presenting products for display during the future time interval, as taught by Lakhani, to the teachings of Hendlin would have yielded predictable results because the level of ordinary skill in the art demonstrated by the references applied shows the ability to incorporate such references into similar non-transitory computer readable mediums. Further, adding the modifications of generating a gradient boosting machine regression model configured to generate a predicted amount of interactions; training at training intervals based on the product feed, the gradient boosting machine regression model by: inputting the aggregated product data into the gradient boosting machine regression model; providing at least one of the product feed or the product feed click count to the gradient boosting machine regression model; updating one or more weights in the gradient boosting machine regression; and optimizing the trained gradient boosting machine regression model by applying Bayesian optimization to one or more hyperparameters associated with the trained gradient boosting machine regression model; generating, with the optimized gradient boosting machine regression model, the predicted amount of interactions; the predicted amount of interactions corresponding to a future time interval; a predicted amount of interactions; and presenting products for display during the future time interval, as taught by Lakhani, into the non-transitory computer readable medium of Hendlin would have been recognized by those of ordinary skill in the art as resulting in an improved non-transitory computer readable medium that would recommend an assortment of items based on estimated demands (Lakhani, [0021]).

Hendlin in view of Lakhani does not explicitly teach receiving the capacity constraint from the second system; and the capacity constraint comprising a numerical limit based on a memory constraint of the second system.
Abuomar, however, teaches rules associated with a capacity (i.e. Col. 10 Ln. 9-16), including the known technique of receiving the capacity constraint from the second system (Abuomar, see at least: “the rule data 122 may not include rules associated with a particular entity identifier 110 … The other rules may output to the entity 108 to solicit user input confirming or rejecting the rules [i.e. receiving the capacity constraint from the second system]” Col. 6 Ln. 53-59 and “A rule module 120 associated with the processing server 106 may determine one or more rules associated with the received data objects 104. In some implementations, the rule module 120 may access rule data 122 [i.e. from the second system]” Col. 6 Ln. 23-26 Examiner notes that entities 108 are the system and the persistence server 106 is the second system); and
the known technique of the capacity constraint comprising a numerical limit based on a memory constraint of the second system (Abuomar, see at least: “FIG. 3 depicts a scenario 300 illustrating a method for determining one or more rules 210 associated with a data object 104 and providing the data object 104 to a data store 102 indicated by the outcome of the rule(s) 210. At 302, a data object 104 may be received by a persistence server 106. In some implementations, the data object 104 may be provided by an entity 108, such as a service, a user, or another type of process or device. In other implementations, the data object 104 may be accessed by the persistence server 106. For example, the data object 104 may be stored in association with the persistence server 106 or in association with a data store 102 or computing device in communication with the persistence server 106. The data object 104 may include data object characteristics 204. For example, the data object characteristics 204 may include physical characteristics of the data object 104, such as a size of the data object (e.g., 2.3 MB) [i.e. the capacity constraint comprising a numerical limit based on a memory constraint of the second system]” Col. 9 Ln. 15-31 and “FIG. 3 depicts an example rule 210 that reads: “If (Size(DataObject)>500 MB OR SalesRank(DataObject)>50) {Store in DataStore1} else {Store in DataStore2}”. The example rule 210 includes three expressions 214 and two threshold values 216. The first expression 214 describes a relationship between a data object characteristic 204 (the size of the data object 204) and a threshold value 216 (500 MB) [i.e. the capacity constraint comprising a numerical limit based on a memory constraint of the second system]” Col. 10 Ln. 9-16 and “Each data object 104 may be stored in a particular data store 102 based on the characteristics of the data object 104 and on the data store characteristics 114 associated with the data stores 102. Data store characteristics 114 may include configurations associated with an associated data store 102, such as the access speed associated with stored data objects 104, a financial cost associated with using or accessing the data store 102, a computational cost associated with use or access of the data store 102, security features associated with the data store 102, storage capacity of the data store 102 [i.e. of the second system], restrictions regarding sizes or types of data objects 104 that may be stored in the data store 102, and so forth” Col. 5 Ln. 42-53 Examiner notes that entities 108 are the system and the persistence server 106 is the second system). These known techniques are applicable to the non-transitory computer readable medium of Hendlin in view of Lakhani as they both share characteristics and capabilities, namely, they are directed to rules associated with a capacity.
It would have been recognized that applying the known techniques of receiving the capacity constraint from the second system; and the capacity constraint comprising a numerical limit based on a memory constraint of the second system, as taught by Abuomar, to the teachings of Hendlin in view of Lakhani would have yielded predictable results because the level of ordinary skill in the art demonstrated by the references applied shows the ability to incorporate such references into similar non-transitory computer readable mediums. Further, adding the modifications of receiving the capacity constraint from the second system; and the capacity constraint comprising a numerical limit based on a memory constraint of the second system, as taught by Abuomar, into the non-transitory computer readable medium of Hendlin in view of Lakhani would have been recognized by those of ordinary skill in the art as resulting in an improved non-transitory computer readable medium that would determine rules associated with the received data objects (Abuomar, Col. 6 Ln. 23-26).

Response to Arguments
Rejections under 35 U.S.C. §101
Applicant argues that the claims are patent-eligible as they do not recite a judicially recognizable exception or grouping of abstract ideas enumerated under Step 2A, Prong One as the current claims are similar to DDR as “determining a capacity constraint associated with a second system configured to display information associated with the one or more products and receiving the capacity constraint from the second system, wherein the capacity constraint is based on a memory constraint of the second system" are particular to transmissions over the internet, as a memory constraint that is received from a system is necessarily rooted in computer technology. See Specification, [0078]. (Remarks, pages 10-11).
Examiner respectfully disagrees. In DDR, the claims overcome a problem or propose a solution “specifically arising in the realm of computer [technology]” DDR Holdings, 773 F.3d at 1257. Unlike the technical problem and technical solution of DDR, optimizing a set of data based on capacity requirements does not address a technical problem that is rooted in technology and is not a technical solution. Accordingly, the claims are ineligible.

Applicant further argues that the claims are integrated into a practical application under Prong Two. The claims reflect improvements in memory consumption for computer systems. Claim 1 recites "determining a capacity constraint associated with a second system configured to display information associated with the one or more products and receiving the capacity constraint from the second system, wherein the capacity constraint is based on a memory constraint of the second system ... selecting, based on the predicted amount of interactions, a second product set. .. based on the capacity constraint" (emphasis added). Applicant disagrees with the Office's assertion that "[m]erely utilizing less data to conserve storage and reduce memory usage is not a technical solution as the actual technology is not providing the solution, the reduction of data is." Office Action, 89. Rather, the claimed capacity constraint of the second system constrains the memory used by the system associated with the webpage. See Specification, [0075],[0098]. Thus, the technology itself improves the functionality of the claimed computer system on which the capacity constraint is imposed by reducing memory usage of that computer system (Remarks, page 12).
Examiner respectfully disagrees. Merely providing less data via a capacity constraint does not improve the computer technology itself as the memory usage of any generic computer would be lowered if less data is sent. Accordingly, the claim fail to reflect an improvement in the functioning of a computer or an improvement to another technology or technical field.

Applicant further argues that the claims reflect improvements in machine learning. Applicant disagrees with the Office's assertion that "Unlike Ex parte Desjardins, it is the data that is being improved rather than the machine learning technology itself." Office Action, 89. Rather, similar to Desjardins, Applicant's Specification explains an improvement to machine learning models, as the Specification explains that training models at various intervals (e.g., based on a product feed) can provide various advantages including improving model performance and accuracy. Specification, [0087]. Furthermore, like in Desjardins, this improvement is reflected in the claim, as claim 1 recites "generating, with the machine learning model, a predicted amount of interactions corresponding to at least one product in the first product set, wherein the machine learning model is trained on the historical product data at training intervals based on the product feed." Thus, claim 1 reflects an improvement to the functioning of a machine learning model, and Applicant submits that claim 1 is eligible for reasons analogous to that of Desjardins (Remarks, pages 12-13).
Examiner respectfully disagrees. In Ex parte Desjardins, the specification identified a technical problem; "effectively learn new tasks in succession whilst protecting knowledge about previous tasks." The solution to this technical problem is reflected in the claims as the claims recite "adjust the first values of the plurality of parameters to optimize performance of the machine learning model on the second machine learning task while protecting performance of the machine learning model on the first machine learning task." Unlike Ex parte Desjardins, it is the data that is being improved rather than the machine learning technology itself; changing the training intervals improves the data provided to the machine learning model, it does not improve the machine learning itself or address a problem that is rooted in technology. Accordingly, the claims are not integrated into a practical application.

Applicant further argues that claims, as a whole, recite significantly more as amended claim 1 recites "mapping each product feed key of the second set of products to a product in the second system; presenting the second product set to the second system to update the product feed; and causing the second system to display the updated the product feed." Applicant's Specification describes how using a product feed key can improve mappings between systems for displaying product feeds and enhancing interactions in a webpage. Specification, [0080], [0095]. Accordingly, Applicant submits that the claims provide improvements to internet-based product feeds over conventional methods of selecting products for a comparison-shopping website. Therefore, the recited claim elements add a specific limitation or combination of limitations that are not well-understood, routine, conventional activity in the field, which is indicative that an inventive concept is present. See M.P.E.P. 2106.05(d) (Remarks, pages 13-14).
Examiner respectfully disagrees. Improving mappings between systems for displaying product data and enhancing interactions are not improvements in the functioning of a computer or an improvements to another technology or technical field. The product data being a product feed and the interactions being in a webpage amount to nothing more than mere instructions to implement or apply the abstract idea on a generic computing hardware (or, merely use a computer as a tool to perform an abstract idea) and do no more than mere instructions to apply the exception using generic computer components.
Furthermore, the additional elements are insufficient to integrate the abstract idea into a practical application because the additional elements amount to nothing more than mere instructions to implement or apply the abstract idea on a generic computing hardware (or, merely use a computer as a tool to perform an abstract idea) and do no more than generally link the use of the judicial exception to a particular technological environment or field of use (such as computers or computing networks). Even considered as an ordered combination (as a whole), the additional elements do not add anything significantly more than when considered individually. Additionally, as is described in the MPEP 2106.05(II) (i.e. “Thus, in Step 2B, examiners should: … Re-evaluate any additional element or combination of elements that was considered to be insignificant extra-solution activity per MPEP § 2106.05(g), because if such re-evaluation finds that the element is unconventional or otherwise more than what is well-understood, routine, conventional activity in the field, this finding may indicate that the additional element is no longer considered to be insignificant”), step 2B considers whether additional elements concluded to be insignificant extra-solution activity in Step 2A are more than well-understood, routine, conventional activity in the field. Examiner did not identify any of the additional elements as insignificant extra-solution activity in Step 2A so there weren’t elements to be evaluated in terms of whether they are more than well-understood, routine, conventional activity in the field. Accordingly, the claims do not amount to significantly more than an abstract idea and are ineligible.

Applicant further argues that claims 11 and 20, although of different scope, are allowable for similar reasons. Dependent claims 2-10 and 12-19 are patent-eligible at least by virtue of their dependence from a patent-eligible independent claim as well as the additional features recited therein. Accordingly, Applicant respectfully requests withdrawal of the rejection under 35 U.S.C. § 101 (Remarks, page 14).
Examiner respectfully disagrees. As detailed in response to the arguments above, the independent claims are not eligible. Accordingly, dependent claims 2-10 and 12-19 are ineligible.

Rejections under 35 U.S.C. §103
Applicant argues that Hendlin is silent regarding at least the “associating,” “mapping,” “presenting,” and “causing” claim elements. Hendlin merely discloses "storing product information corresponding to a product feed at the social networking system." Hendlin, [0065]. However, Hendlin fails to teach or suggest mapping a product key of the second set of products to a product in the second system. In addition, Lakhani and Abuomar, alone or in combination with Hendlin, fail to cure such disclosure of Hendlin. Thus, the cited references fail to teach or suggest each element of claim 1 (Remarks, pages 14-15).
Examiner respectfully disagrees. Hendlin discloses that the social networking system (i.e. second system) stores the product information received from the product feed in association which a particular merchant and the stored data being updated (e.g. product names, product styles, product groups, product sales, product prices, and/or other product attributes) by updates from the merchant product feed [i.e. mapping a product key of the second set of products to a product in the second system] (see Hendlin, [0061], [0033], and [0065]). The product data is initially stored at the second system (i.e. a product in the second system) and this stored product data is updated via the product feed from the merchant system. Accordingly, the cited references teach the amended claims.

Applicant further argues that independent claims 11 and 20, although different in scope, recite subject matter similar to that of independent claim 1 and are allowable over the cited references for at least the same reasons. Dependent claims 2-10 and 12-19 depend either directly or indirectly from independent claims 1, 11, and 20 and are allowable at least based on their dependence from allowable base claims (Remarks, page 15).
Examiner respectfully disagrees. As detailed in response to the argument above, independent claim 1 is not allowable. Accordingly, independent claims 11 and 20 and dependent claims 2-10 and 12-19 are not allowable.

Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
-Bathla et al. (US 2024/0354822 A1) teaches handling requests for information related items in a data feed.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to ARIELLE E WEINER whose telephone number is [phone]. The examiner can normally be reached M-F 8:30-5:00.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Maria-Teresa (Marissa) Thein can be reached at 571-272-6764. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.

/ARIELLE E WEINER/ Primary Examiner, Art Unit 3689

Prosecution Timeline

• Sep 01, 2023: Application Filed
• Sep 18, 2025: Non-Final Rejection mailed — §101, §103
• Nov 18, 2025: Interview Requested
• Nov 25, 2025: Examiner Interview Summary
• Nov 25, 2025: Applicant Interview (Telephonic)
• Dec 16, 2025: Response Filed
• Apr 02, 2026: Final Rejection mailed — §101, §103
• Jun 24, 2026: Request for Continued Examination
• Jul 02, 2026: Response after Non-Final Action
• Jul 29, 2026: Non-Final Rejection mailed — §101, §103 (current)