DETAILED ACTION
Status of the Application
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and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114.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 11/18/2025 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 .
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 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 action is a Non-Final Action on the merits in response to the application filed on 11/18/2025.
Claims 1, 10, and 18 have been amended.
Claims 2, 6, 11, 14, 19, 21 have been cancelled.
Claims 1, 3-5, 7-10, 12, 13, 15-18, 20 remain pending in this application.
Response to Amendment
Applicant’s amendments are acknowledged.
The 35 U.S.C. 101 rejections of claims in the previous office action have been maintained.
The 35 U.S.C. 112 rejections of claims in the previous office action are withdrawn in light of applicant’s amendments.
The 35 U.S.C. 103 rejections of claims in the previous office action are withdrawn in light of applicant’s amendments.
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, 3-5, 7-9 are directed towards a system, claims 10, 12, 13, 15-17 are directed towards a method, and claims 18, 20 are directed towards a computer readable medium, all of which are among the statutory categories of invention.
Step 1: This part of the eligibility analysis evaluates whether the claim falls within any statutory category. See MPEP 2106.03. The claim recites at least one step or act, including training of models. Thus, the claim is to a process, which is one of the statutory categories of invention. (Step 1: YES).
Step 2A, Prong One: This part of the eligibility analysis evaluates whether the claim recites a judicial exception. As explained in MPEP 2106.04, subsection II, a claim “recites” a judicial exception when the judicial exception is “set forth” or “described” in the claim.
With respect to claims 1, 3-5, 7-10, 12, 13, 15-18, 20, the independent claims (claims 1, 10, and 18) are directed to managing sales and advertisement data, In independent claim 1, the bolded limitations emphasized below correspond to the abstract ideas of the claimed invention:
Claim 1, A system comprising:
obtain, from a data repository, first item data for a first item that is a low-velocity item;
obtain, from the data repository, second item data for a plurality of second items each of which is a high-velocity item;
generate a plurality of before-sale features based on product descriptions in the first item data and the second item data, wherein each of the plurality of before-sale features characterizes an item before the item is offered for sale;
generate, from the plurality of second items, a ranked list of substitute items for the first item based on the plurality of predicted substitution scores;
store the ranked list of substitute items and their predicted substitution scores in the data repository;
receive, from a server, a substitution request for the first item that is out of stock.
these steps fall within and recite an abstract ideas because they are directed to a method of organizing human activity which includes managing commercial interaction such as advertising, sales activities, or business relations (See MPEP 2106.04(a)(2) II).
If a claim limitation, under its broadest reasonable interpretation, covers commercial interaction, then it falls within the “method of organizing human activity” grouping of abstract ideas. Therefore, If the identified limitation(s) falls within any of the groupings of abstract ideas enumerated in the MPEP 2106, the analysis should proceed to Prong Two. (Step 2A, Prong One: YES).
Step 2A, Prong Two: This part of the eligibility analysis evaluates whether the claim as a whole integrates the recited judicial exception into a practical application of the exception or whether the claim is “directed to” the judicial exception. This evaluation is performed by (1) identifying whether there are any additional elements recited in the claim beyond the judicial exception, and (2) evaluating those additional elements individually and in combination to determine whether the claim as a whole integrates the exception into a practical application. See MPEP 2106.04(d). The claim recites the additional elements of processor, memory, machine learning model, server ( claim 10 processor, memory, machine learning model, server; claim 18 computer-readable medium, processor, memory, machine learning model, server). The claims recite the steps are performed by the processor, memory, machine learning model, server.
The limitations of
a processor; and
a non-transitory memory storing instructions that, when executed, cause the processor to:
train a machine learning model based on features generated from product descriptions for a plurality of high-velocity items and corresponding substitution scores between the plurality of high-velocity items, wherein the training comprises adjusting weights to map the features generated from the product descriptions for the plurality of high-velocity items to the corresponding substitution scores between the plurality of high-velocity items;
compute a plurality of predicted substitution scores based at least in part by applying the trained machine learning model to the plurality of before- sale features, wherein each of the plurality of predicted substitution scores represents a degree of likelihood to substitute the first item with a corresponding second item of the plurality of second items when the first item is out of stock;
transmit a substitution response including the ranked list of substitute items to the server to trigger displaying, next to a first advertisement for the first item on a webpage, a second advertisement for at least one substitute item in the ranked list of substitute items.
are mere data processing and outputting recited at a high level of generality, and thus are insignificant extra-solution activity. See MPEP 2106.05(g) (“whether the limitation is significant”). In addition, all uses of the recited judicial exceptions require such data gathering and output, and, as such, these limitations do not impose any meaningful limits on the claim. These limitations amount to necessary data gathering and outputting. See MPEP 2106.05.
Further, the limitations are recited as being performed by m processor, memory, machine learning model, server. The processor, memory, machine learning model, server are recited at a high level of generality. In limitation (a), the machine learning model is used as a tool to perform the generic computer function of receiving data. See MPEP 2106.05(f). The machine learning model is used to perform an abstract idea, as discussed above in Step 2A, Prong One, such that it amounts to no more than mere instructions to apply the exception using a generic computer. See MPEP 2106.05(f). Additionally, claim 1 recites machine learning model. The general use of a machine learning technique does not provide a meaningful limitation to transform the abstract idea into a practical application.
Even when viewed in combination, these additional elements do not integrate the recited judicial exception into a practical application (Step 2A, Prong Two: NO), and the claim is directed to the judicial exception. (Step 2A: YES).
Step 2B: This part of the eligibility analysis evaluates whether the claim as a whole amounts to significantly more than the recited exception i.e., whether any additional element, or combination of additional elements, adds an inventive concept to the claim. See MPEP 2106.05. As explained with respect to Step 2A, Prong Two, the additional elements are the processor, memory, machine learning model, server. The additional elements were found to be insignificant extra-solution activity in Step 2A, Prong Two, because they were determined to be insignificant limitations as necessary processor, memory, machine learning model, server. Then, the machine learning techniques recited in the claim are disclosed at a high-level of generality (see at least Specification [0009 “the computing device trains a machine learning process based on the features to learn a mapping of the features to output data characterizing predicted substitution scores. The computing device also stores configuration parameters associated with the trained machine learning process in a data repository. 0038“Once trained, item substitution determination computing device 102 may store the machine learning model parameters (e.g., hyperparameters, configuration settings, weights, etc.) associated with the machine learning process within database 116. As such, during inference, item substitution determination computing device 102 may obtain the parameters from database 116, configure the machine learning model with or based on the obtained parameters, and execute the machine learning model accordingly.””]) and does not amount to significantly more than the abstract idea.
However, a conclusion that an additional element is insignificant extra solution activity in Step 2A, Prong Two should be re-evaluated in Step 2B. See MPEP 2106.05, subsection I.A. At Step 2B, the evaluation of the insignificant extra-solution activity consideration takes into account whether or not the extra-solution activity is well understood, routine, and conventional in the field. See MPEP 2106.05(g). As discussed in Step 2A, Prong Two above, the recitations of
a processor; and
a non-transitory memory storing instructions that, when executed, cause the processor to:
train a machine learning model based on features generated from product descriptions for a plurality of high-velocity items and corresponding substitution scores between the plurality of high-velocity items, wherein the training comprises adjusting weights to map the features generated from the product descriptions for the plurality of high-velocity items to the corresponding substitution scores between the plurality of high-velocity items;
compute a plurality of predicted substitution scores based at least in part by applying the trained machine learning model to the plurality of before- sale features, wherein each of the plurality of predicted substitution scores represents a degree of likelihood to substitute the first item with a corresponding second item of the plurality of second items when the first item is out of stock;
transmit a substitution response including the ranked list of substitute items to the server to trigger displaying, next to a first advertisement for the first item on a webpage, a second advertisement for at least one substitute item in the ranked list of substitute items..
are recited at a high level of generality. These elements amount to transmitting data and are well understood, routine, conventional activity. See MPEP 2106.05(d), subsection II. 10 As discussed in Step 2A, Prong Two above, the recitation of a processor, memory, machine learning model, server to perform limitations amounts to no more than mere instructions to apply the exception using a generic computer component. Even when considered in combination, these additional elements represent mere instructions to implement an abstract idea or other exception on a computer and insignificant extra-solution activity, which do not provide an inventive concept. (Step 2B: NO).
Dependent claims 3-5, 7-9, 12, 13, 15-17, 20 do not contain any new additional elements. Rather, these claims offer further descriptive limitations of elements found in the independent claims. In this case, the claims are rejected for the same reasons at step 2a, prong one; step 2a, prong 2; and step 2b. Thus, the claim is not patent eligible.
Regarding the dependent claims, dependent claims 3-5, 7, 8, 12, 13, 15, 16, 20 recite machine learning model. The dependent claims 3-5, 7-9, 12, 13, 15-17, 20 recite limitations that are not technological in nature and merely limits the abstract idea to a particular environment. Claims 3-5, 7-9, 12, 13, 15-17, 20 recites processor, memory, machine learning model, server which are considered an insignificant extra-solution activities of collecting and analyzing data; see MPEP 2106.05(g). Claims 3-5, 7-9, 12, 13, 15-17, 20 recites processor, memory, machine learning model, server, which merely recites an instruction to apply the abstract idea using a generic computer component; MPEP 2106.05(f). Additionally, claims 3-5, 7-9, 12, 13, 15-17, 20 recite steps that further narrow the abstract idea. No additional elements are disclosed in the dependent claims that were not considered in independent claims 1, 10, and 18. Therefore claims 3-5, 7-9, 12, 13, 15-17, 20 do not provide meaningful limitations to transform the abstract idea into a patent eligible application of the abstract idea such that the claims amount to significantly more than the abstract idea itself.
Reasons for Removing the Prior Art Rejection
The rejections under 35 U.S.C. 103 as to claims 1, 3-5, 7-10, 12, 13, 15-18, 20 are removed in light of Applicant's claims and remarks of 11/18/2025, which are deemed persuasive as to independent claim 1. The reasons for withdrawal of the rejections under 35 U.S.C. 103 can be found at the following claim limitations of 11/18/2025 at claim 1 as follows:
Claim 1, A system comprising:
a processor; and
a non-transitory memory storing instructions that, when executed, cause the processor to:
train a machine learning model based on features generated from product descriptions for a plurality of high-velocity items and corresponding substitution scores between the plurality of high-velocity items, wherein the training comprises adjusting weights to map the features generated from the product descriptions for the plurality of high-velocity items to the corresponding substitution scores between the plurality of high-velocity items;
obtain, from a data repository, first item data for a first item that is a low-velocity item;
obtain, from the data repository, second item data for a plurality of second items each of which is a high-velocity item;
generate a plurality of before-sale features based on product descriptions in the first item data and the second item data, wherein each of the plurality of before-sale features characterizes an item before the item is offered for sale;
compute a plurality of predicted substitution scores based at least in part by applying the trained machine learning model to the plurality of before- sale features, wherein each of the plurality of predicted substitution scores represents a degree of likelihood to substitute the first item with a corresponding second item of the plurality of second items when the first item is out of stock;
generate, from the plurality of second items, a ranked list of substitute items for the first item based on the plurality of predicted substitution scores;
store the ranked list of substitute items and their predicted substitution scores in the data repository;
receive, from a server, a substitution request for the first item that is out of stock; and
transmit a substitution response including the ranked list of substitute items to the server to trigger displaying, next to a first advertisement for the first item on a webpage, a second advertisement for at least one substitute item in the ranked list of substitute items.
Applicant’s Remarks of 11/18/2025 at pg. 11-13 as follows:
“Nowhere, however, does Pande disclose training a machine learning model "based on features generated from product descriptions for a plurality high-velocity items and corresponding substitution scores between the high-velocity items," let alone where the training "cause[s] the machine learning model to learn weights to map product description features to predicted substitution scores," as recited in unamended claim 1. In other words, while in Pande the generated graph may include edge weights, nowhere does Pande teach training a machine learning model based on features generated from product descriptions for a plurality high-velocity items and corresponding substitution scores between those same high-velocity items. Indeed, and to the extent the Final Office Action is associating the graph weights with the claimed substitution scores, which Applicant maintains is improper, nonetheless Pande teaches that the graph weights are calculated based on past customer views for each corresponding pair of items. See, e.g., Pande, para. [0084].
Nonetheless, and although Applicant disagrees with the rejection, merely to expedite prosecution, Applicant has amended the independent claims. For example, independent claim 1 now recites, among other things:
train a machine learning model based on features generated
from product descriptions for a plurality of high-velocity items and
corresponding substitution scores between the plurality of high- velocity items, wherein the training comprises adjusting weights to
map the features generated from the product descriptions for the
plurality of high-velocity items to the corresponding substitution
scores between the plurality of high-velocity items;
obtain, from a data repository, first item data for a first item that is
a low-velocity item;
obtain, from the data repository, second item data for a plurality of
second items each of which is a high-velocity item;
generate a plurality of before-sale features based on product
descriptions in the first item data and the second item data, wherein
each of the plurality of before-sale features characterizes an item before
the item is offered for sale;
compute a plurality of predicted substitution scores based at
least in part by applying the trained machine learning model to the
plurality of before-sale features, wherein each of the plurality of
predicted substitution scores represents a degree of likelihood to
substitute the first item with a corresponding second item of the plurality
of second items when the first item is out of stock;
generate, from the plurality of second items, a ranked list of
substitute items for the first item based on the plurality of predicted
substitution scores;
store the ranked list of substitute items and their predicted
substitution scores in the data repository;
receive, from a server, a substitution request for the first item that
is out of stock; and
transmit a substitution response including the ranked list of
substitute items to the server to trigger displaying, next to a first
advertisement for the first item on a webpage, a second advertisement
for at least one substitute item in the ranked list of substitute items.
(Emphases added). Although different in scope, independent claims 10 and 18 recite similar features. The cited portions of Usrey, Pande, and Roesbery, alone or in any proper combination, fail to teach or suggest these features.
As such, and in accordance with recommendations made in the Final Office Action, amended claim 1 positively recites the training of the machine learning model and, specifically, recites "train[ing] a machine learning model based on features generated from product descriptions for a plurality of high-velocity items and corresponding substitution scores between the plurality of high-velocity items, wherein the training comprises adjusting weights to map the features generated from the product descriptions for the plurality of high-velocity items to the corresponding substitution scores between the plurality of high-velocity items." See Final Office Action, p. 29. As such, as claimed, the machine learning model is trained with 1) features generated from product descriptions for a plurality of high-velocity items; and 2) features generated from substitution scores between the plurality of high-velocity items. As noted above, however, the cited portions of Pande relate to a graph with edge weights corresponding to a degree of substitutability between items, where the graph weights are calculated based on past customer views for each corresponding pair of items. Nowhere, however, does Pande teach the above quoted features of claim 1, which are similarly recited in claims 10 and 18. Moreover, none of the cited portions of Usrey and Roesbery, alone or in combination cure these deficiencies of Pande.”
This applies to independent claims 10 and 18 as these claims includes the same feature of claim 1.
Page 9 of 13
Response to Arguments
Applicant’s arguments filed 11/18/2025 have been fully considered but they are not persuasive. Applicant’s arguments will be addressed hereinbelow in the order in which they appear in the response filed 11/18/2025.
Regarding the 35 U.S.C. 101 rejection, at pg. 14-17 Applicant argues with respect to claims at issue are not directed to an abstract idea
In response to the 35 USC § 101 claim rejection argument, the Examiner respectfully disagrees. The Examiner did consider each claim and every limitation both individually and as a whole, since the grounds of rejection clearly indicates that an abstract idea has been identified from elements recited in the claims. Using the two-part analysis, the Office has determined there are no elements, in the claim sufficient enough to ensure that the claims amounts to significantly more than the abstract idea itself. As recited, the claims are directed towards:
Claim 1, A system comprising:
a processor; and
a non-transitory memory storing instructions that, when executed, cause the processor to:
train a machine learning model based on features generated from product descriptions for a plurality of high-velocity items and corresponding substitution scores between the plurality of high-velocity items, wherein the training comprises adjusting weights to map the features generated from the product descriptions for the plurality of high-velocity items to the corresponding substitution scores between the plurality of high-velocity items;
obtain, from a data repository, first item data for a first item that is a low-velocity item;
obtain, from the data repository, second item data for a plurality of second items each of which is a high-velocity item;
generate a plurality of before-sale features based on product descriptions in the first item data and the second item data, wherein each of the plurality of before-sale features characterizes an item before the item is offered for sale;
compute a plurality of predicted substitution scores based at least in part by applying the trained machine learning model to the plurality of before- sale features, wherein each of the plurality of predicted substitution scores represents a degree of likelihood to substitute the first item with a corresponding second item of the plurality of second items when the first item is out of stock;
generate, from the plurality of second items, a ranked list of substitute items for the first item based on the plurality of predicted substitution scores;
store the ranked list of substitute items and their predicted substitution scores in the data repository;
receive, from a server, a substitution request for the first item that is out of stock; and
transmit a substitution response including the ranked list of substitute items to the server to trigger displaying, next to a first advertisement for the first item on a webpage, a second advertisement for at least one substitute item in the ranked list of substitute items.
The claim(s) does not include additional elements that are sufficient to amount to significantly more than the judicial exception because the computer as recited is a generic computer component that performs functions.
Examiner finds the claim recite concepts which are now described in the 2019 PEG as certain methods of organizing human activity. In particular the claims recites limitations for managing sales and advertisement data, which constitutes methods related to commercial interactions such as advertising, sales activities, or business relations which are still considered an abstract idea under the 2019 PEG. The trained machine learning is comprised of generic elements to perform an existing business process. Examiner finds the claims recite mere instructions to implement the abstract idea on a computer and uses the computer as a tool to perform the abstract idea without reciting any improvements to a technology, technological process or computer-related technology or a machine learning model.
Regarding, the steps at pg. 16 that Applicant points to as practical application/significantly more are merely narrowing the abstract idea to a particular technological environment, which has been found to be ineffective to render an abstract idea eligible. Furthermore, the Examiner respectfully disagrees because the steps of:
“Instead, independent claim 1 requires the execution of each of the claimed features, including the use of a trained machine learning model to generate the ranked list of substitute items for the first item, and the transmission of a substitution response including the ranked list of substitute items to a server to trigger the displaying, next to a first advertisement for the first item on a webpage, of a second advertisement for at least one substitute item in the ranked list of substitute items, as recited in claim 1, and similarly recited in claims 10 and 18.”
and arguments at pg. 17 seems to describe a “particular way” of managing sales and advertisement data are part of the abstract idea. “ The Applicant is basically relying on the trained machine learning model as integrating the abstract idea into a practical application but those trained machine learning model are not being improved. The general use of machine learning techniques does not provide a meaningful limitation to transform the abstract idea into a practical application. As such, In regards to Appeal 2024-000567, Ex Parte Desjardins, the instant claims are not similar to Ex Parte Desjardins, Examiner finds the Board determined the improvements in Desjardins to be directed to addressing problems arising in the context of a technical improvements to machine learning systems, which overcome a problem specifically arising in the realm of AI and machine learning inventions. There is no similar technological problem or solution here. The claims discloses the defining of machine learning models at a high-level of generality, without incorporating any updating (i.e. training) limitations. Therefore, currently, the machine learning recited in the claims is solely used a tool to perform the instructions of the abstract idea. There is no similar technological problem or solution here, as the current claims are just using typically known actions/steps of a machine learning model and no improvements. Additionally, the Examiner did consider “the Kim Memo” and all examination falls in line with the memo.
At pg. 16, the Applicant argues that “Here, the claims provide a technical solution rooted in computer technology to address problems with existing item recommendation systems, and do not merely link a technical environment of computers to "fundamental economic or commercial practices."”
In response, the Examiner respectfully disagree, the Examiner would like to direct the Applicant to, DDR Holdings, LLC v. Hotels.com, L.P. case or Example 36; where the DDR invention and Example 36 were necessarily rooted in computer technology. Applicant’s claims are not necessarily rooted in computer technology. Rather, Applicant’s invention aims to solve a business problem—for accessing and managing sales and advertisement data—rather than a technological one. Indeed, the computing elements and trained machine learning model recited in Applicant’s claims are merely generic in nature and are insufficient to shift the focus or thrust of Applicant 's invention. Moreover, Applicant’s claims resemble a web/server administrator scheme for solving a business solution with merely generic computer implementation. Additionally, the Examiner would like to point the Applicant to the 2019 PEG, in which managing sales and advertisement data will fall under. The 2019 PEG which states:
Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea - see MPEP 2106.05(f).
Adding insignificant extra-solution activity to the judicial exception - see MPEP 2106.05(g)
Generally linking the use of the judicial exception to a particular technological environment or field of use – see MPEP 2106.05(h)
Therefore, the additional elements do not integrate into a practical application and not similar to Example 36.
At pg. 16, Applicant argues that claims are similar to Trading Technologies and Core Wireless.
In response, Examiner respectfully disagrees. In the instant case the examiner asserts that the fact pattern in Trading Technologies does not uniquely match the fact pattern in the instant case. For example, In Trading Technologies "The court distinguished this system from the routine or conventional use of computers or the Internet, and concluded that the specific structure and concordant functionality of the graphical user interface are removed from abstract ideas, as compared to conventional computer implementations of known procedures. Thus the court held that the criteria of Alice Step 2 were also met."
The instant claims are not drawn to a particular and non-conventional arrangement of elements of a graphical user interface. As such, the examiner asserts that the fact patterns of the instant case and Trading Technologies are different.
Regarding Core Wireless, the Examiner notes that Core Wireless was found eligible not only because the claims are directed to a particular manner of summarizing and presenting information in electronic devices, but also because of the improvements to the user interface for electronic devices. Examiner finds the advancements disclosed in Core Wireless are not comparable the present claims. In particular, the combination of structural elements recited in the present claim are used to perform generic functions that are well-understood, routine and conventional (e.g. trained machine learning model)
Examiner finds, the present claims are more similar to concepts identified as abstract by the courts, such as sending, directing, monitoring receipt of, and accumulating records about information in Two-Way Media Ltd. v. Comcast Cable Communications.
Regarding the 35 U.S.C. 103 rejection, Applicant’s arguments with respect to claims has been considered but are moot in view of the new grounds of rejection.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
* WO 2019083714 A1 teaching System for calculating competitive interrelationships in item-pair
* Pande et al, Substitution Techniques for Grocery Fulfillment and Assortment Optimization Using Product Graphs, InProceedings of the 16th International Workshop on Mining and Learning with Graphs MLG, 2020
* Thangali et al, Price investment using prescriptive analytics and optimization in retail, InProceedings of the 26th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, p3136-p3144, Aug 23 2020
* Thangali et al, Robust Price Optimization in Retail, Proceedings of the 38th International Conference on Machine Learning, PMLR 139, 2021.
* Bouyarmane; KarimUS 11734242 B1 teaching Architecture For Resolution Of Inconsistent Item Identifiers In A Global Catalog
* US 20190121867 A1 hereinafter Misra teaching- “applying a machine learning model to the plurality of features generated based on the product descriptions, to compute predicted substitution scores between the first item and each of the plurality of second items”
(Misra ¶ [0151] 2nd sentence and ¶ [0154] 2nd sentence: the propensity score for every item-pair in the variety substitute item-pair cluster > 0, indicating the items in each item-pair may qualify as variety substitutes for each other, depending on their individual scores and/or rankings. Specifically, see
Misra ¶ [0158] 2nd-3rd sentences: multi-stage cluster component determines competitive interrelationships between items in a category to identify traditional substitutes and variety substitutes of selected items. The system analyzes the overall profile of an item-pair via culling data from diverse sources and use machine learning techniques. Additionally, see ¶ [0165] 3rd clause after colon, noting in addition a second-stage clustering component that performs a filtration cluster operation on the first cluster of substitute item-pairs using an item description similarity variable with the POS data and the item attribute data to generate… a second cluster of substitute item-pairs. Misra ¶ [0084] The second-stage cluster component uses the item description similarity, as the feature in the K-means(k=2) clustering algorithm in the second-stage clustering. This feature proxies for the different attributes present in the items. Substitute items are more similar in terms of the attributes, as compared to non-substitutes. Substitute item description similarity has higher values. After formation of the two clusters of item-pairs, profiling is performed in terms of item description similarity to detect the cluster with the finer set of substitute item-pairs, which is identified as the second cluster of substitute item-pairs 328. In some examples, the second cluster of substitute item-pairs 328 is merged with the first cluster of substitute item-pairs 320. See another example at ¶ [0116]);
- “wherein each of the plurality of features characterizes an item [in the future]
(Misra ¶ [0104] 4th sentence: The higher rank may also indicate a higher likelihood that item C will be purchased together with item A by consumers in the future).
- “wherein each of the predicted substitution scores represents a degree of likelihood to substitute the first item with a corresponding second item”;
(Misra ¶ [0091] 2nd-3rd sentences: For traditional substitute items, the purchase of one item in an item-pair nullifies the purchase probability of the other item in the item-pair because both items in the item-pair serve the same need/function. Therefore, the lift value for a traditional pair is likely to be lower because traditional substitute items are not bought in the same transaction. Similarly, ¶ [0121] 4th-5th sentences.
Misra ¶ [0108] measure of association variable 602 is calculated measure of association between the two items in a given item-pair. The measure of association is a transformation of yulesQ, a commonly used metric to understand product association. As the measure of association for a given item-pair increases, the chances [or likelihood] that the item-pair is a substitute item-pair also increases. In other words, there is a correlation between the measure of association between items and likelihood of the items being classified as substitute items. Misra ¶ [0109] 2nd sentence: Given an item-pair, the multi-stage cluster component calculates how purchase of a first item pi changes the odds [or likelihood] of purchase of a second item pj. Misra ¶ [0130] 5th sentence: As the percentage of same-basket purchases of two items increases, the [or likelihood] probability that these two items are variety substitutes also increases)
- “generate, from the plurality of second items, a ranked list of substitute items for the first item based on the predicted substitution scores”
(Misra ¶ [0054] ranking component 122 generates a ranking for each variety substitute of the selected item in the cluster of variety item-pairs. A variety substitute is an item in at least one item-pair in the cluster of variety substitute item-pairs. The ranking is generated based on the calculated propensity score for each item-pair in the cluster of variety item-pair substitutes. ¶ [0102] Fig. 5 illustrates a multi-stage clustering analysis result 500 with item ranks. The result 500 includes an identification of a set of one or more variety item-pairs 502 and/or an identification of a set of one or more traditional substitute item-pairs 504. Also see Misra ¶ [0178] ranking, by a ranking component, an item associated with a selected item in each item-pair in the second cluster of substitute item-pairs, the ranking generated based on a propensity score assigned to each item-pair by a scoring component; ¶ [0179] identifying, by a selection component, an item associated with at least one item-pair having a lowest score-based rank and a selection-rate below a threshold rate for a predetermined time-range for removal from inventory; ¶ [0180] ranking, by a ranking component, each item associated with at least one item-pair in the cluster of traditional substitute item-pairs based on the propensity score assigned to each item-pair by a scoring component; ¶ [0181] identifying an item having a lowest rank and a selection-rate below a threshold rate for a predetermined time-range for removal from inventory, wherein the identified item is associated with at least one traditional substitute item)
- “store the predicted substitution scores in a data repository” (Misra ¶ [0056] 1st-2nd sentences: system 100 include data storage device 132 storing data, such as item substitute ranking(s) 130, the set of interrelationship variables 126, item data 134 associated with the items, the propensity score(s)128, etc.).
* US 20080249658 A1 ¶ [0169] 1st sentence: The second exemplary record of the resolution rules database provides that a substitute product offer should be provided to a customer when …(2) the average actual item velocity for other products in the same category as the initially requested product is less than the average ideal item velocity for other products in the same category…
¶ [0172] Executing the exemplary rules of FIG. 7 in accordance with the exemplary data provided by the databases of FIGS. 4-6, it may be determined that, should Soda X jam in the inventory storage and dispensing mechanism 170 after a customer deposited $0.65 (Soda X's retail price per FIG. 5A), a substitute product offer for Sodas Y or Z is appropriate as a resolution because: (1) Sodas Y and Z are in the same product category ("soda") as Soda X, the initially requested product; (2) the average actual item velocity for Sodas Y and Z is 0.5/day, which is less than the average ideal item velocity for Sodas Y and Z (2.5/day); and (3) the total coin inventory ($3.40) less the deposited amount ($0.65) is less than that which may be needed to make change for future customers within the fill period. Specifically, the transaction database 120 of FIGS. 4A and 4B indicates that the average change dispensed per day is .about.$0.35. Assuming, for this example, that there are 13 days remaining in the fill period, $4.55 in coin inventory may be needed to make change for customers throughout the remainder of the fill period, assuming that the historic transaction patterns in the transaction database 120 (e.g., transaction velocity, change due, etc.) are indicative of future transaction patterns. Thus, the amount of change anticipatorily needed to make change for the remainder of the fill period ($4.55) is greater than the total coin inventory less the amount deposited ($2.75).
* US 20220129900 A1 ¶ [0061] Trust score contour 260, also includes a second trust measure of SIM tenure, which may relate, for example, to recent occurrences of events affecting the SIM of a mobile communications device. Thus, a mobile communications device subscriber who, in a recent period of time, has replaced a SIM less frequently than indicated by historical norm 216, for example, may be assigned a second trust measure of SIM tenure, as indicated at SIM replacement point 245. In an example, historical norm 216 may indicate a SIM replacement velocity of one replacement per year, while recent SIM replacement point 245 may indicate a velocity of 0.5 SIM replacements per year. Accordingly, SIM replacement point 245 is positioned, for example, along subscriber identity module tenure axis 215 at a location closer to trust space origin 201 than historical norm 216.
* US 20160304282 A1 ¶ [0021] In the retail industry, the term “fast-moving” products is used to denote products that are sold frequently, such as the better-known brands of soft drinks and coffee. The term “slow-moving” products as used in the retail industry refers to products that are sold much less frequently. All the products of the total product range of products to be stored in the product warehouse can be classed as being of the “slow-moving” type or the “fast-moving” type, although it is also quite conceivable within the framework of the present invention that all the products of an order are determined to be products of the “slow-moving” type, or at least of a first type, i.e. that all the products listed in the order are exclusively of the “slow-moving” type, or at least of a first type. The division of products into products of the “slow-moving” type and products of the “fast-moving” type can for example be made on the basis of the “Pareto” principle, which is well-known in the art, according to which a division is made, for example, in which 20% of the products are determined to be “fast-moving” products, which products occur in about 80% of the orders. The remaining 80% of the products is thus of the “slow-moving” type. With such a division of products into “slow-moving” products and “fast-moving” products, and—when collecting products of an order—by supplying a product of the “fast-moving” type which is not already available in the module directly from a product carrier, usually a pallet, in the pallet warehouse, irrespective of the fact whether said product is available in another module, the number of product conveying operations is strongly reduced, because the products of the “fast-moving” type that are stored in the various modules will anyhow be shortly collected for another order that is allocated to the module in question and need not be supplied from elsewhere, therefore. On the other hand, when products of the “slow-moving” type which have not been allocated to another order yet, i.e. which are available, are by contrast collected from other modules, if possible, such products will on the one hand not remain in the product warehouse for an undesirably long period of time, whilst on the other hand it will not be necessary to unload an entire product carrier for such a low-demand product whilst products of the kind in question are still present in the product warehouse, which would lead to an unnecessary increase of the total number of such products stored in the storage warehouse.
Fig.3 below and ¶ [0059] As already explained in the foregoing, products for the retail trade can usually be divided into products of a “slow-moving” type, hereinafter called “slow-movers” and products of a “fast-moving” type, hereinafter called “fast-movers”. ¶ [0063] Calculations have shown that moving fast-movers between modules is less efficient than—if a fast-mover is needed for an order and the product in question is not present in the module to which the order has been allocated—unloading the product in question directly from a product carrier carrying a number of the respective products being needed, and storing the other products from said product carriers in the product warehouse 5, distributed over the modules. ¶ [0064] Consequently, the method according to the invention preferably comprises moving product i from a product carrier from the pallet warehouse to the module in accordance with step S8, while products listed in the order are being collected in module 1, if a product of a “fast-moving” type, i.e. a fast-mover, listed in the order is not already present in module 1 (Step S3 and Step S4, NO to both questions), wherein furthermore, according to step S7, the product in question is allocated to the order. Then it is verified again, in accordance with step S2, whether the order is already complete. The product carrier is preferably unloaded in its entirety, with the further products from the carrier being distributed over the total number of modules in the product warehouse 5 and thus being available for further orders.
PNG
media_image1.png
506
705
media_image1.png
Greyscale
Fig.3 of US 20160304282 A1
* Grob et al, US 20230169442 A1 hereinafter Grob teaching the following:
- “obtain first item data for a first item and second item data for a plurality of second items”
(Grob ¶ [0041] 2nd - 3rd sentences: First, at training time, at operation 802, data regarding a plurality of sample products is accessed. The data includes four or more attribute types and, for each product in the plurality of sample products, a value for each of the four or more attribute types);
- “generate a plurality of features based on the first item data and the second item data”
(Grob [0018] 4th-7th sentences 1st machine learning algorithm 116 trains 1st machine-learned model 114 to take master data relating to products and turn it into multidimensional vectors. Each vector contains dimensions, with each dimension related to a different feature of the data, pertaining to a single product. A feature represents a particular type field of data. Since data here pertains to products, each feature is different attribute of a product, with values for each feature representing values for the attributes);
- “map the plurality of features to output data characterizing a predicted substitution score”;
(Grob ¶ [0020] 2nd-7th sentences: user specify a product discontinued by manufacturer, and similarity service 124 then query data structure 118 to determine products similar to discontinued product. This may be performed by similarity service 124 identifying the set of coordinates associated with the discontinued product then locating sets of coordinates within some threshold geometric distance to discontinued product in 3D space, specified by data structure 118. Those sets of coordinates within the threshold geometric distance then represent matching products for the discontinued product, which allows assortment manager to then order those matching products to make up for loss of discontinued product. In some example embodiments, user may specify hypothetical or dream product to have similarity service 124 identify close matches to. In this way, similarity service 124 act as recommendation service. A questionnaire may be presented to users that allow the users to identify features of such a hypothetical or dream product.
Grob ¶ [0033] 2nd-8th sentences: The embedding identifies a set of coordinates (called a vector) in a multidimensional space to each parsed sentence. One or more similarity measures between coordinates may then be used to identify the similarity between corresponding sentences. One example metric is cosine similarity. In cosine similarity, the cosine of the angle between two vectors is computed. The cosine similarity of two sentences will range from 0 to 1. If the cosine similarity is 1, it means the two vectors have the same orientation and thus are identical. Values closer to 0 indicate less similarity. ¶ [0034]-¶ [0035] In example embodiment BERT are used to encode sentences into embeddings in sentence encoder 204. It should be noted that BERT is merely used in some example embodiments, and in others another language model can be used in lieu of BERT. FIG. 3 is a block diagram illustrating BERT, in accordance with an example embodiment. BERT uses transformer layer(s) 300 to encode the input sentence to embedding. Each transformer layer is defined as follows:
PNG
media_image2.png
217
559
media_image2.png
Greyscale
where hn-1 is output of previous transformer layer. Here, BERT model with 8 transformer layers is used, and the output sentence embedding zsent is defined as the meanpooling result of the last transformer layer's output. ¶ [0041] training comprises feeding the vectors for the sample products into the 1st machine learning algorithm, and 1st machine learning algorithm, in response to the feeding, learning an embedding in 3D space for each vector, with the embedding representing a set of parameters to apply to input vector to map the input vector to set of coordinates in 3D space. ¶ [0043] At operation 816, it is determined how similar 1st product in plurality of products is to 2nd product in the plurality of products based on a geometric distance between the set of coordinates corresponding to the embedding for the 1st product and the set of coordinates corresponding to the embedding for 2nd product, as stored in the data structure) “and”
- “store the predicted substitution score in a data repository”
(Grob ¶ [0042] 5th sentence to ¶ [0044]: at operation 814, the embedding for each product are stored in a data structure. At operation 816, it is determined how similar a first product in the plurality of products is to a second product in the plurality of products based on a geometric distance between the set of coordinates corresponding to the embedding for the first product and the set of coordinates corresponding to the embedding for the second product, as stored in the data structure).
- “mapping the plurality of features to the output data comprises establishing a Sentence Bidirectional Encoder Representations from Transformers (SBERT) model”
(Grob ¶ [0034]-¶ [0035] BERT are used to encode sentences into embeddings in sentence encoder 204. Fig.3 illustrate BERT which uses transformer layer(s) 300 to encode the input sentence to embedding. Each transformer layer is defined as:
PNG
media_image2.png
217
559
media_image2.png
Greyscale
where hn-1 is output of previous transformer layer. Here, BERT model with 8 transformer layers is used, and the output sentence embedding zsent is defined as meanpooling result of the last transformer layer's output).
“wherein the computing device is configured to”:
- “obtain substitution scores between the plurality of third items” (Grob ¶[0033] 3rd-8th sentences similarity measures between coordinates may then be used to identify similarity between corresponding sentences. One example metric is cosine similarity. In cosine similarity, cosine of the angle between two vectors is computed. The cosine similarity of two sentences will range from 0 to 1. If cosine similarity is 1, it means the 2 vectors have same orientation and thus are identical. Values closer to 0 indicate less similarity. Other measures of similarity may be used, in lieu of or in addition to cosine similarity, such as Euclidean distance)
- “generate respective features based on the product descriptions and the substitution scores”; (Grob mid-¶ [0028] master data may include textual descriptions of the products, and these textual descriptions may be mined for additional features of the product not identified as distinct features in the metadata ¶ [0034] 4th sentence: BERT uses transformer layer(s) 300 to encode the input sentence to embedding and
- “apply the SBERT model to the respective features to learn the mapping of the plurality of features to the output data” (Grob ¶ [0034]-¶ [0035] BERT are used to encode sentences into embeddings in sentence encoder 204. Fig.3 illustrate BERT which uses transformer layer(s) 300 to encode the input sentence to embedding. Each transformer layer is defined as:
PNG
media_image3.png
196
500
media_image3.png
Greyscale
where hn-1 is output of previous transformer layer. Here, BERT model with 8 transformer layers is used, and the output sentence embedding zsent is defined as meanpooling result of the last transformer layer's output).
Any inquiry concerning this communication or earlier communications from the examiner should be directed to UCHE BYRD whose telephone number is (571)272-3113. The examiner can normally be reached Mon.-Fri..
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, Patricia Munson can be reached at (571) 270-5396. 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.
/UCHE BYRD/Examiner, Art Unit 3624
Prosecution Insights