DETAILED ACTION
Status of Claims
• The following is a non-final, first office action in response to the communication filed 04/01/2024.
• Claims 1-20 are currently pending and have been examined.
Priority
The examiner acknowledges that the instant application is a continuation of US Patent No. 11948179, filed 01/31/2021.
Information Disclosure Statement
Information Disclosure Statement received 09/05/2024 has been reviewed and considered.
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 .
Double Patenting
The nonstatutory double patenting rejection is based on a judicially created doctrine grounded in public policy (a policy reflected in the statute) so as to prevent the unjustified or improper timewise extension of the “right to exclude” granted by a patent and to prevent possible harassment by multiple assignees. A nonstatutory double patenting rejection is appropriate where the conflicting claims are not identical, but at least one examined application claim is not patentably distinct from the reference claim(s) because the examined application claim is either anticipated by, or would have been obvious over, the reference claim(s). See, e.g., In re Berg, 140 F.3d 1428, 46 USPQ2d 1226 (Fed. Cir. 1998); In re Goodman, 11 F.3d 1046, 29 USPQ2d 2010 (Fed. Cir. 1993); In re Longi, 759 F.2d 887, 225 USPQ 645 (Fed. Cir. 1985); In re Van Ornum, 686 F.2d 937, 214 USPQ 761 (CCPA 1982); In re Vogel, 422 F.2d 438, 164 USPQ 619 (CCPA 1970); In re Thorington, 418 F.2d 528, 163 USPQ 644 (CCPA 1969).
A timely filed terminal disclaimer in compliance with 37 CFR 1.321(c) or 1.321(d) may be used to overcome an actual or provisional rejection based on nonstatutory double patenting provided the reference application or patent either is shown to be commonly owned with the examined application, or claims an invention made as a result of activities undertaken within the scope of a joint research agreement. See MPEP § 717.02 for applications subject to examination under the first inventor to file provisions of the AIA as explained in MPEP § 2159. See MPEP §§ 706.02(l)(1) - 706.02(l)(3) for applications not subject to examination under the first inventor to file provisions of the AIA . A terminal disclaimer must be signed in compliance with 37 CFR 1.321(b).
The USPTO Internet website contains terminal disclaimer forms which may be used. Please visit www.uspto.gov/patent/patents-forms. The filing date of the application in which the form is filed determines what form (e.g., PTO/SB/25, PTO/SB/26, PTO/AIA /25, or PTO/AIA /26) should be used. A web-based eTerminal Disclaimer may be filled out completely online using web-screens. An eTerminal Disclaimer that meets all requirements is auto-processed and approved immediately upon submission. For more information about eTerminal Disclaimers, refer to www.uspto.gov/patents/process/file/efs/guidance/eTD-info-I.jsp.
Claims 1-20 are rejected on the ground of nonstatutory double patenting as being unpatentable over claims 1-2, and 4-11 of U.S. Patent No. 11948179 B2 (referred to as “‘179”).
‘179 teaches:
Claim 1
‘179 [Claim 1]
A system comprising: one or more processors; and one or more non-transitory computer-readable media storing computing instructions that, when executed on the one or more processors, cause the one or more processors to perform operations comprising:
A system comprising: one or more processors; and one or more non-transitory computer-readable media storing computing instructions that, when executed on the one or more processors, cause the one or more processors to perform operations comprising:
generating, using a trained machine-learning model, personalized product-type metrics for a user based on historic activity of the user and product-type pairs in an item taxonomy;
training a machine-learning model to learn item-level embedding Gaussian distributions for items, a user embedding, and product-type embedding Gaussian mixture distributions based on co-purchase item pairs in historic activity data of a user and product-type pairs in an item taxonomy…generating personalized product-type metrics for the user based at least in part on the user embedding for the user and the product-type embedding Gaussian mixture distributions
determining, using the personalized product-type metrics for the user, top product types for the user based on an anchor item;
receiving a request for personalized complementary item recommendations for an anchor item and the user, the items comprising the anchor item…personalized product-type metrics for the user…determining top product types based at least in part on personalized product-type complementarity metrics generated using the personalized product-type metrics and cosine similarity measurements
determining a set of first items associated with the top product types;
generating a set of first items of the items associated with the top product types;
ranking each item in the set of first items using at least item-level embedding Gaussian distributions for (i) the anchor item and (ii) for each item in the set of first items; and
receiving a request for personalized complementary item recommendations for an anchor item and the user, the items comprising the anchor item…ranking each respective item in the set of first items based on a respective item-to-item complementarity metric for each respective item generated using an item-level embedding Gaussian distribution of the item-level embedding Gaussian distributions for the anchor item and a respective item-level embedding Gaussian distribution item-level embedding Gaussian distributions for each respective item
selecting, based on the ranking, a set of top items from the set of first items to be personalized complementary item recommendations for the user based on the anchor item.
receiving a request for personalized complementary item recommendations for an anchor item and the user, the items comprising the anchor item…selecting a set of top items as the personalized complementary item recommendations based on the ranking
Claim 2
‘179 [Claim 10]
wherein the historic activity of the user comprises add-to-cart data of the user.
wherein the historic activity data of the user comprises add-to-cart sequence data of the user.
Claim 3
‘179
wherein: determining the set of first items associated with the top product types comprises: generating, by a cosine similarity measurement between a mean vector of the item-level embedding Gaussian distribution for the anchor item and a respective mean vector of an item-level embedding Gaussian distribution for each respective item, a respective item-to-item complementarity metric for each respective item; and
[Claim 1] – ranking each respective item in the set of first items based on a respective item-to-item complementarity metric for each respective item generated using an item-level embedding Gaussian distribution of the item-level embedding Gaussian distributions for the anchor item and a respective item-level embedding Gaussian distribution item-level embedding Gaussian distributions for each respective item
[Claim 2] – wherein the respective item-to-item complementarity metric for each respective item is generated using a cosine similarity measurement between a mean vector of the item-level embedding Gaussian distribution for the anchor item and a respective mean vector of the respective item-level embedding Gaussian distribution for each respective item
ranking each item in the set of first items comprises: ranking, by the respective item-to-item complementarity metric, each item in the set of first items.
[Claim 1] – ranking each respective item in the set of first items based on a respective item-to-item complementarity metric for each respective item
Claim 4
‘179
wherein the operations further comprise: training a machine-learning model to learn item-level embedding Gaussian distributions, a user embedding, and product-type embedding Gaussian mixture distributions for items based on (i) co-purchase item pairs in the historic activity of the user and (ii) the product-type pairs in the item taxonomy; and
[Claim 1] – training a machine-learning model to learn item-level embedding Gaussian distributions for items, a user embedding, and product-type embedding Gaussian mixture distributions based on co-purchase item pairs in historic activity data of a user and product-type pairs in an item taxonomy
for an item pair of the co-purchase item pairs: retrieving respective Gaussian distribution representations of each item of the item pair
[Claim 4] – for an item pair of the co-purchase item pairs: retrieving respective Gaussian distribution representations of each item of the item pair
determining, using a probability product kernel, an item complementarity metric between each item of the item pair; and determining a user-to-item preference metric for both items in the item pair, wherein the trained machine-learning model comprises the machine-learning model, as trained.
[Claim 4] – determining an item complementarity metric between each item of the item pair using a probability product kernel; and determining a user-to-item preference metric for both items in the item pair
[Claim 1] – training a machine-learning model
Claim 5
‘179 [Claim 5]
wherein the operations further comprise: generating loss functions based on (i) the item complementarity metric and (ii) a user-to-item preference metric
for the item pair of the co-purchase item pairs: generating loss functions based on the item complementarity metric and the user-to-item preference metric.
Claim 6
‘179 [Claim 7]
wherein training the machine-learning model further comprises at least one of: optimizing the loss functions using gradient descent; or updating the item-level embedding Gaussian distributions, the user embedding, and the product-type embedding Gaussian mixture distributions based on hyperparameters of the loss functions
wherein training the item-level embedding Gaussian distributions for items, the user embedding, and the product-type embedding Gaussian mixture distributions further comprises: optimizing loss functions using gradient descent; and updating the item-level embedding Gaussian distributions, the user embedding, and the product-type embedding Gaussian mixture distributions based on hyperparameters of the loss functions, as optimized
Claim 7
‘179 [Claim 9]
wherein the item-level embedding Gaussian distributions are configured to model a correction for popular co-purchased items
wherein the item-level embedding Gaussian distributions model a correction for popular co-purchased items
Claim 8
‘179 [Claim 8]
wherein the product-type embedding Gaussian mixture distributions are configured to model topic disambiguation among multi-topic product types
wherein the product-type embedding Gaussian mixture distributions model topic disambiguation among multi-topic product types
Claim 9
‘179 [Claim 6]
wherein the operations further comprise: training a product-type embedding Gaussian mixture distributions for items for a product-type pair of the product-type pairs, wherein the training comprises: retrieving respective Gaussian mixture representations of each product type of the product-type pairs; determining a non-personalized product-type complementarity metric at a global level between each product type of the product-type pairs using a probability product kernel with a non-personalized component weight; and determining a personalized product-type complementarity metric at an individual level between each product type of the product-type pair using a probability product kernel with a personalized component weight
wherein training the product-type embedding Gaussian mixture distributions for the items further comprises, for an product-type pair of the product-type pairs: retrieving respective Gaussian mixture representations of each product type of the product-type pair; determining a non-personalized product-type complementarity metric at a global level between each product type of the product-type pair using a probability product kernel with a non-personalized component weight; determining a personalized product-type complementarity metric at an individual level between each product type of the product-type pair using the probability product kernel with a personalized component weight
Claim 10
‘179 [Claim 9]
wherein training the product-type embedding Gaussian mixture distributions further comprises: generating loss functions based on the non-personalized product-type complementarity metric and the personalized product-type complementarity metric
wherein training the product-type embedding Gaussian mixture distributions for the items further comprises, for an product-type pair of the product-type pairs: …generating loss functions based on the non-personalized product-type complementarity metric and the personalized product-type complementarity metric.
In regards to claim 11, claim 11 is directed to a method. Claim 11 recites limitations that are substantially parallel in nature to those addressed above for claim 1 which is directed towards a system. The system of ‘179 discloses the limitations of claim 1 as noted above. ‘179 further discloses a method implemented via execution of computing instructions configured to run on one or more processors and stored at one or more non-transitory computer-readable media, the method comprising (‘179: [Claim 11]). Claim 11 is therefore rejected for the reasons set forth above in claim 1 and in this paragraph.
In regards to claims 12-20, all the limitations in method claims 12-20 are closely parallel to the limitations of system claims 2-10 analyzed above and rejected on the same bases.
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 without significantly more. The claims recite an abstract idea. The judicial exception is not integrated into a practical application. The claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception.
First, it is determined whether the claims are directed to a statutory category of invention. See MPEP 2106.03(II). In the instant case, claims 1-10 are directed to a machine, and claims 11-20 are directed to a process. Therefore, claims 1-20 are directed to statutory subject matter under Step 1 of the Alice/Mayo test (Step 1: YES).
The claims are then analyzed to determine if the claims are directed to a judicial exception. See MPEP 2106.04. In determining whether the claims are directed to a judicial exception, the claims are analyzed to evaluate whether the claims recite a judicial exception (Prong 1 of Step 2A), as well as analyzed to evaluate whether the claims recite additional elements that integrate the judicial exception into a practical application of the judicial exception (Prong 2 of Step 2A). See MPEP 2106.04.
Taking claim 1 as representative, claim 1 recites at least the following limitations that are believed to recite an abstract idea:
generating, using a model, personalized product-type metrics for a user based on historic activity of the user and product-type pairs in an item taxonomy;
determining, using the personalized product-type metrics for the user, top product types for the user based on an anchor item;
determining a set of first items associated with the top product types;
ranking each item in the set of first items using at least item-level embedding Gaussian distributions for (i) the anchor item and (ii) for each item in the set of first items; and
selecting, based on the ranking, a set of top items from the set of first items to be personalized complementary item recommendations for the user based on the anchor item.
The above limitations recite the concept of analyzing data to provide recommendations. These limitations, under their broadest reasonable interpretation, fall within the “Certain Methods of Organizing Human Activity” grouping of abstract ideas, enumerated in the MPEP, in that they recite commercial or legal interactions such as advertising, marketing, or sales activities or behaviors. Specifically, providing recommendations is a marketing and sales activity. This is further illustrated in paragraph [0003] of the Specification, describing the invention relates to shopping. Independent claim 11 recites similar limitations as claim 1 and as such, claim 10 falls within the same identified grouping of abstract ideas. Accordingly, under Prong One of Step 2A of the Alice/Mayo test, claims 1 and 11 recite an abstract idea (Step 2A, Prong One: YES).
Under Prong Two of Step 2A of the MPEP, claims 1 and 11 recite additional elements, such as a system comprising: one or more processors; and one or more non-transitory computer-readable media storing computing instructions that, when executed on the one or more processors, cause the one or more processors to perform operations, a trained machine-learning model, and execution of computing instructions configured to run on one or more processors and stored at one or more non-transitory computer-readable media. These additional elements are described at a high level in Applicant’s specification without any meaningful detail about their structure or configuration. As such, these computer-related limitations are not found to be sufficient to integrate the abstract idea into a practical application. Although these additional computer-related elements are recited, claims 1 and 11 merely invoke such additional elements as a tool to perform the abstract idea. Implementing an abstract idea on a generic computer is not indicative of integration into a practical application. Similar to the limitations of Alice, claims 1 and 11 merely recite a commonplace business method (i.e., analyzing data to provide recommendations) being applied on a general purpose computer. See MPEP 2106.05(f). Furthermore, claims 1 and 11 generally link the use of the abstract idea to a particular technological environment or field of use. The courts have identified various examples of limitations as merely indicating a field of use/technological environment in which to apply the abstract idea, such as specifying that the abstract idea of monitoring audit log data relates to transactions or activities that are executed in a computer environment, because this requirement merely limits the claims to the computer field, i.e., to execution on a generic computer (see FairWarning v. Iatric Sys.). Likewise, claims 1 and 11 specifying that the abstract idea of analyzing data to provide recommendations is executed in a computer environment merely indicates a field of use in which to apply the abstract idea because this requirement merely limits the claims to the computer field, i.e., to execution on a generic computer. As such, under Prong Two of Step 2A of the MPEP, when considered both individually and as a whole, the limitations of claims 1 and 11 are not indicative of integration into a practical application (Step 2A, Prong Two: NO).
Since claims 1 and 11 recite an abstract idea and fail to integrate the abstract idea into a practical application, claims 1 and 11 are “directed to” an abstract idea (Step 2A: YES).
Next, under Step 2B, the claims are analyzed to determine if there are additional claim limitations that individually, or as an ordered combination, ensure that the claim amounts to significantly more than the abstract idea. See MPEP 2106.05. The instant claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception for at least the following reasons.
Returning to independent claims 1 and 11, these claims recite additional elements, such as a system comprising: one or more processors; and one or more non-transitory computer-readable media storing computing instructions that, when executed on the one or more processors, cause the one or more processors to perform operations, a trained machine-learning model, and execution of computing instructions configured to run on one or more processors and stored at one or more non-transitory computer-readable media. As discussed above with respect to Prong Two of Step 2A, although additional computer-related elements are recited, the claims merely invoke such additional elements as a tool to perform the abstract idea. See MPEP 2106.05(f). Moreover, the limitations of claims 1 and 11 are manual processes, e.g., receiving information, analyzing information, etc. The courts have indicated that mere automation of manual processes is not sufficient to show an improvement in computer-functionality (see MPEP 2106.05(a)(I)). Furthermore, as discussed above with respect to Prong Two of Step 2A, claims 1 and 11 merely recite the additional elements in order to further define the field of use of the abstract idea, therein attempting to generally link the use of the abstract idea to a particular technological environment, such as the Internet or computing networks (see Ultramercial, Inc. v. Hulu, LLC. (Fed. Cir. 2014); Bilski v. Kappos (2010); MPEP 2106.05(h)). Similar to FairWarning v. Iatric Sys., claims specifying that the abstract idea of analyzing data to provide recommendations is executed in a computer environment merely indicates a field of use in which to apply the abstract idea because this requirement merely limits the claim to the computer field, i.e., to execution on a generic computer.
Even when considered as an ordered combination, the additional elements do not add anything that is not already present when they are considered individually. In Alice Corp., the Court considered the additional elements “as an ordered combination,” and determined that “the computer components…‘[a]dd nothing…that is not already present when the steps are considered separately’ and simply recite intermediated settlement as performed by a generic computer.” Id. (citing Mayo, 566 U.S. at 79, 101 USPQ2d at 1972). Similarly, viewed as a whole, claims 1 and 11 simply convey the abstract idea itself facilitated by generic computing components. Therefore, under Step 2B of the Alice/Mayo test, there are no meaningful limitations in claims 1 and 11 that transform the judicial exception into a patent eligible application such that the claims amount to significantly more than the judicial exception itself (Step 2B: NO).
Dependent claims 2-10 and 12-20, when analyzed as a whole, are held to be patent ineligible under 35 U.S.C. 101 because they recite an abstract idea, are not integrated into a practical application, and do not add “significantly more” to the abstract idea. More specifically, dependent claims 2-10 and 12-20 further fall within the “Certain Methods of Organizing Human Activity” grouping of abstract ideas, enumerated in the MPEP, in that they further recite commercial or legal interactions such as advertising, marketing, or sales activities or behaviors and managing personal behavior or relationships or interactions between people. Dependent claims 2-3, 5-8, 12-13, and 15-18 fail to identify additional elements and as such, are not indicative of integration into a practical application. Dependent claims 4, 9-10, 14, and 19-20 further identify additional elements, such as training a machine-learning model, and training product-type embedding Gaussian mixture distributions. Similar to discussion above the with respect to Prong Two of Step 2A, although additional computer-related elements are recited, the claims merely invoke such additional elements as a tool to perform the abstract idea. See MPEP 2106.05(f). As such, under Step 2A, dependent claims 2-10 and 12-20 are “directed to” an abstract idea. Similar to the discussion above with respect to claims 1 and 11, dependent claims 2-10 and 12-20 analyzed individually and as an ordered combination, invoke such additional elements as a tool to perform the abstract idea and merely indicate a field of use in which to apply the abstract idea because this requirement merely limits the claims to the computer field, i.e., to execution on a generic computer, and therefore, do not amount to significantly more than the abstract idea itself. See MPEP 2106.05(f)(2). Accordingly, under the Alice/Mayo test, claims 1-20 are ineligible.
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 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.
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 set forth in Graham v. John Deere Co., 383 U.S. 1, 148 USPQ 459 (1966), that are applied 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 1-3 and 11-13 are rejected under 35 U.S.C. 103 as being unpatentable over Al Jadda et al. (US 20200334734 A1), hereafter Al Jadda, in view of Canal et al. (US 20220129709 A1), hereinafter Canal.
In regards to claim 1, Al Jadda discloses a system comprising: one or more processors; and one or more non-transitory computer-readable media storing computing instructions that, when executed on the one or more processors, cause the one or more processors to perform operations comprising: (Al Jadda: [0004])
generating, using a trained machine-learning model, personalized product-type metrics for a user based on historic activity of the user and product-type pairs in an item taxonomy (Al Jadda: [0004] – “determining, based on co-purchase data, a percentage for each of the plurality of taxonomies that any product in the first taxonomy has been co-purchased with any product of each of the respective plurality of taxonomies”; [0015] – “recommendations may be provided in conjunction with the…purchase of a so-called ‘anchor product’ or ‘anchor item’”; [0060] – “The neural network may have been trained on manually-created sets of similar products”);
determining, using the personalized product-type metrics for the user, top product types for the user based on an anchor item (Al Jadda: [0055] – “A predefined number of closest taxonomies may be determined for the anchor product taxonomy, in some embodiments. For example, five (5) closest taxonomies may be determined based on co-purchase data”; [0003] – “determining, by the one or more processors, a second taxonomy closest to the first taxonomy. The second taxonomy may be associated with a group of products, the first taxonomy and the second taxonomy have at least a common highest taxonomy level, and the determination that the second taxonomy is closest to the first taxonomy may be made at least in part based on co-purchase data indicating that the anchor product and at least one product in the group of products are purchased together more often than products associated with other taxonomies are purchased with the anchor product”);
determining a set of first items associated with the top product types (Al Jadda: [0003] – “determining, by the one or more processors, a most similar product to the anchor product from the group of products of the second taxonomy”; [0059] – “for a given taxonomy from the closest taxonomies determined in step 40, each product within the taxonomy is evaluated for its similarity to the anchor product”; [0061] – “A cosine similarity function may be applied to the respective vectors of each pairing of anchor product and second product to determine which second product(s) from the plurality of products is/are most similar to the anchor product within the taxonomy”; [0055] and Fig. 4 – “A predefined number of closest taxonomies may be determined for the anchor product taxonomy, in some embodiments. For example, five (5) closest taxonomies may be determined based on co-purchase data…In the example of FIG. 4, each of the products 48, 50, 52, 54, and 56 may belong to a respective one of the five taxonomies that have products that are most likely to be co-purchased with products of the taxonomy associated with the anchor product 46”);
ranking each item in the set of first items using at least item-level embedding for (i) the anchor item and (ii) for each item in the set of first items (Al Jadda: [0046] – “Candidate pairs (e.g., potential products that may be recommended as in a product collection with the anchor product) may be generated, and those candidate pairs may be ranked”; [0060-0061] – “inputting data respective of the anchor product and of each of the plurality of products in the taxonomy into a trained neural network to generate a vector respective of each product… A cosine similarity function may be applied to the respective vectors of each pairing of anchor product and second product to determine which second product(s) from the plurality of products is/are most similar to the anchor product within the taxonomy…a biLSTM may be used in various embodiments to determine similar products to an anchor product…generate text embeddings for products…The embedding outputs F(x.sub.i), F(x.sub.j) may be joined by a metric function at the top. Then, a cosine similarity…may be used”; [0059] – “for a given taxonomy from the closest taxonomies determined in step 40, each product within the taxonomy is evaluated for its similarity to the anchor product”); and
selecting, based on the ranking, a set of top items from the set of first items to be personalized complementary item recommendations for the user based on the anchor item (Al Jadda: [0046] – “Candidate pairs (e.g., potential products that may be recommended as in a product collection with the anchor product) may be generated, and those candidate pairs may be ranked”; [0066] – “Compute the cosine similarity sim for the pair (a, p.sub.i) (e.g., using co-purchase data) end Rank all similarities sim(a, p.sub.i) computed; Add to the list l the product p.sub.i for which the similarity sim(a, p.sub.i) has the highest value”; [0055] and Fig. 4 – “each of the products 48, 50, 52, 54, and 56 may belong to a respective one of the five taxonomies that have products that are most likely to be co-purchased with products of the taxonomy associated with the anchor product 46”).
Al Jadda further discloses an analysis including generating embeddings (Al Jadda: [0061]), yet Al Jadda does not explicitly disclose the analysis using Gaussian distributions.
However, Canal teaches a similar recommender system (Canal: [0026]), including
the analysis using Gaussian distributions (Canal: [0077] – “Imposing Gaussian distributions on inter-object distances…approach to modeling uncertainty in ordinal embeddings”).
It would have been obvious to one of ordinary skill in the art at the time the invention was filed to have included the distribution of Canal in the system of Al Jadda because Al Jadda already discloses an analysis and Canal is merely demonstrating how the analysis may occur. Additionally, it would have been obvious to have included the analysis using Gaussian distributions as taught by Canal because Gaussian distributions are well-known and the use of it in a recommender system would have enabled approximation of the distance posterior with a fixed batch of samples from a simple distribution (Canal: [0077]).
In regards to claim 2, Al Jadda/Canal teaches the system of claim 1. Al Jadda further discloses wherein the historic activity of the user comprises add-to-cart data of the user (Al Jadda: [0031] – “receiving a selection of an anchor product from a user…selection may be…a user action to add the anchor product to the user's shopping cart on the website”).
In regards to claim 3, Al Jadda/Canal teaches the system of claim 1. Al Jadda further discloses determining the set of first items associated with the top product types comprises: generating, by a cosine similarity measurement between a vector of the item-level embedding for the anchor item and a respective vector of an item-level embedding for each respective item, a respective item-to-item complementarity metric for each respective item (Al Jadda: [0003] – “determining, by the one or more processors, a most similar product to the anchor product from the group of products of the second taxonomy”; [0059] – “for a given taxonomy from the closest taxonomies determined in step 40, each product within the taxonomy is evaluated for its similarity to the anchor product”; [0061] – “A cosine similarity function may be applied to the respective vectors of each pairing of anchor product and second product to determine which second product(s) from the plurality of products is/are most similar to the anchor product within the taxonomy”; [0055] and Fig. 4 – “A predefined number of closest taxonomies may be determined for the anchor product taxonomy, in some embodiments. For example, five (5) closest taxonomies may be determined based on co-purchase data…In the example of FIG. 4, each of the products 48, 50, 52, 54, and 56 may belong to a respective one of the five taxonomies that have products that are most likely to be co-purchased with products of the taxonomy associated with the anchor product 46”); and
ranking each item in the set of first items comprises: ranking, by the respective item-to-item complementarity metric, each item in the set of first items (Al Jadda: [0046] – “Candidate pairs (e.g., potential products that may be recommended as in a product collection with the anchor product) may be generated, and those candidate pairs may be ranked”; [0060-0061] – “inputting data respective of the anchor product and of each of the plurality of products in the taxonomy into a trained neural network to generate a vector respective of each product… A cosine similarity function may be applied to the respective vectors of each pairing of anchor product and second product to determine which second product(s) from the plurality of products is/are most similar to the anchor product within the taxonomy…a biLSTM may be used in various embodiments to determine similar products to an anchor product…generate text embeddings for products…The embedding outputs F(x.sub.i), F(x.sub.j) may be joined by a metric function at the top. Then, a cosine similarity…may be used”; [0059] – “for a given taxonomy from the closest taxonomies determined in step 40, each product within the taxonomy is evaluated for its similarity to the anchor product”),
yet Al Jadda does not explicitly disclose information comprising a mean; and the analysis including Gaussian distribution.
However, Canal further teaches
information comprising a mean (Canal: [0052] – “selecting pairs uniformly at random and user preferences are estimated as the posterior mean”); and
the analysis including Gaussian distribution (Canal: [0077] – “Imposing Gaussian distributions on inter-object distances…approach to modeling uncertainty in ordinal embeddings”).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine Canal with Al Jadda for the reasons identified above with respect to claim 1.
In regards to claim 11, claim 11 is directed to a method. Claim 11 recites limitations that are substantially parallel in nature to those addressed above for claim 1 which is directed towards a system. The combined system of Al Jadda/Canal teaches the limitations of claim 1 as noted above. Al Jadda further discloses a method implemented via execution of computing instructions configured to run on one or more processors and stored at one or more non-transitory computer-readable media, the method comprising: (Al Jadda: [0004]). Claim 11 is therefore rejected for the reasons set forth above in claim 1 and in this paragraph.
In regards to claims 12-13, all the limitations in method claims 12-13 are closely parallel to the limitations of system claims 2-3 analyzed above and rejected on the same bases.
Claims 4, 7-8, 14, and 17-18 are rejected under 35 U.S.C. 103 as being unpatentable over Al Jadda , in view of Canal, in view of Segev et al. (US 10148680 B1), hereinafter Segev.
In regards to claim 4, Al Jadda/Canal teaches the system of claim 1. Al Jadda further discloses wherein the operations further comprise: training a machine-learning model to learn item-level embedding, a user embedding, and product-type embedding for items based on (i) co-purchase item pairs in the historic activity of the user and (ii) the product-type pairs in the item taxonomy (Al Jadda: [0003] – “determining, by the one or more processors, a most similar product to the anchor product from the group of products of the second taxonomy”; [0059] – “for a given taxonomy from the closest taxonomies determined in step 40, each product within the taxonomy is evaluated for its similarity to the anchor product”; [0061] – “A cosine similarity function may be applied to the respective vectors of each pairing of anchor product and second product to determine which second product(s) from the plurality of products is/are most similar to the anchor product within the taxonomy”; [0055] and Fig. 4 – “A predefined number of closest taxonomies may be determined for the anchor product taxonomy, in some embodiments. For example, five (5) closest taxonomies may be determined based on co-purchase data…In the example of FIG. 4, each of the products 48, 50, 52, 54, and 56 may belong to a respective one of the five taxonomies that have products that are most likely to be co-purchased with products of the taxonomy associated with the anchor product 46”; [0068] – “using aggregated co-purchase transactional data”); and
for an item pair of the co-purchase item pairs: retrieving respective representations of each item of the item pair; determining an item complementarity metric between each item of the item pair; and determining a user-to-item preference metric for both items in the item pair, wherein the trained machine-learning model comprises the machine-learning model, as trained (Al Jadda: [0003] – “determining, by the one or more processors, a most similar product to the anchor product from the group of products of the second taxonomy”; [0059] – “for a given taxonomy from the closest taxonomies determined in step 40, each product within the taxonomy is evaluated for its similarity to the anchor product”; [0061] – “A cosine similarity function may be applied to the respective vectors of each pairing of anchor product and second product to determine which second product(s) from the plurality of products is/are most similar to the anchor product within the taxonomy”; [0055] and Fig. 4 – “A predefined number of closest taxonomies may be determined for the anchor product taxonomy, in some embodiments. For example, five (5) closest taxonomies may be determined based on co-purchase data…In the example of FIG. 4, each of the products 48, 50, 52, 54, and 56 may belong to a respective one of the five taxonomies that have products that are most likely to be co-purchased with products of the taxonomy associated with the anchor product 46”; [0068] – “using aggregated co-purchase transactional data”; [0060] – “The neural network may have been trained on manually-created sets of similar products”),
yet Al Jadda does not explicitly disclose the analysis including Gaussian distributions, Gaussian mixture distributions and Gaussian distribution; and using a probability product kernel.
However, Canal further teaches
the analysis including Gaussian distributions and Gaussian distribution (Canal: [0077] – “Imposing Gaussian distributions on inter-object distances…approach to modeling uncertainty in ordinal embeddings”); and
using a probability product kernel (Canal: [0080] – “the probabilistic multi-dimensional scaling (MDS) technique used by CKL”; [0006] – “Crowd Kernel Learning (CKL)”).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine Canal with Al Jadda for the reasons identified above with respect to claim 1.
Additionally, Segev teaches a similar ranking system (Segev: Col. 9, Ln. 38-40), including
the analysis using Gaussian mixture distributions (Segev: Col. 20, Ln. 62-63 – “Applying a Gaussian mixture fit to the histogram computed in the precedent stage”).
It would have been obvious to one of ordinary skill in the art at the time the invention was filed to have included the Gaussian mixture of Segev in the system of Al Jadda/Canal because Al Jadda/Canal already discloses an analysis and Segev is merely demonstrating how the analysis may occur. Additionally, it would have been obvious to have included the analysis using Gaussian mixture distributions as taught by Segev because Gaussian mixture distributions are well-known and the use of it in a recommender system would have improved efficiency (Segev: Col. 26, Ln. 40-41).
In regards to claim 7, Al Jadda/Canal/Segev teaches the system of claim 4. Al Jadda further discloses wherein the item-level embedding are configured to model a correction for popular co-purchased items (Al Jadda: [0045-0046] and Fig. 5 – “FIG. 5 shows the loyalty to five popular brands computed from 99,754 transactions through an online retailer that involved exactly 4 products that belonged to the bathroom category for the year 2018…a similar conclusion—end-consumers are not necessarily brand loyal when co-purchasing products. Accordingly, product collection recommendations that may be biased toward recommending products of the same brand together may not be optimal… Accordingly, described herein are various embodiments for identifying relationships between products that indicate they are good fits for a collection of products”; see also [0060-0061]).
yet Al Jadda does not explicitly disclose the analysis including Gaussian distributions.
Additionally, Segev teaches a similar ranking system (Segev: Col. 9, Ln. 38-40), including
the analysis using Gaussian mixture distributions (Segev: Col. 20, Ln. 62-63 – “Applying a Gaussian mixture fit to the histogram computed in the precedent stage”).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine Segev with Al Jadda/Canal for the reasons identified above with respect to claim 1.
In regards to claim 8, Al Jadda/Canal/Segev teaches the system of claim 4. Al Jadda further discloses wherein the product-type embedding are configured to model topic disambiguation among multi-topic product types (Al Jadda: [0046] – “Candidate pairs (e.g., potential products that may be recommended as in a product collection with the anchor product) may be generated, and those candidate pairs may be ranked”; [0060-0061] – “inputting data respective of the anchor product and of each of the plurality of products in the taxonomy into a trained neural network to generate a vector respective of each product… A cosine similarity function may be applied to the respective vectors of each pairing of anchor product and second product to determine which second product(s) from the plurality of products is/are most similar to the anchor product within the taxonomy…a biLSTM may be used in various embodiments to determine similar products to an anchor product…generate text embeddings for products…The embedding outputs F(x.sub.i), F(x.sub.j) may be joined by a metric function at the top. Then, a cosine similarity…may be used”).
yet Al Jadda does not explicitly disclose the analysis including Gaussian distributions.
However, Canal further teaches
the analysis including Gaussian distributions (Canal: [0077] – “Imposing Gaussian distributions on inter-object distances…approach to modeling uncertainty in ordinal embeddings”).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine Canal with Al Jadda for the reasons identified above with respect to claim 4.
In regards to claims 14 and 17-18, all the limitations in method claims 14 and 17-18 are closely parallel to the limitations of system claims 4 and 7-8 analyzed above and rejected on the same bases.
Claims 5-6 and 15-16 are rejected under 35 U.S.C. 103 as being unpatentable over Al Jadda , in view of Canal, in view of Segev, in view of David Wei et al. (US 20180047071 A1), hereinafter David Wei.
In regards to claim 5, Al Jadda/Canal/Segev teaches the system of claim 4. Al Jadda further discloses wherein the operations further comprise: (i) the item complementarity metric and (ii) a user-to-item preference metric (Al Jadda: [0003] – “determining, by the one or more processors, a second taxonomy closest to the first taxonomy. The second taxonomy may be associated with a group of products, the first taxonomy and the second taxonomy have at least a common highest taxonomy level, and the determination that the second taxonomy is closest to the first taxonomy may be made at least in part based on co-purchase data indicating that the anchor product and at least one product in the group of products are purchased together more often than products associated with other taxonomies are purchased with the anchor product”; [0061] – “A cosine similarity function may be applied to the respective vectors of each pairing of anchor product and second product to determine which second product(s) from the plurality of products is/are most similar to the anchor product within the taxonomy”; [0055] and Fig. 4 – “A predefined number of closest taxonomies may be determined for the anchor product taxonomy, in some embodiments. For example, five (5) closest taxonomies may be determined based on co-purchase data…In the example of FIG. 4, each of the products 48, 50, 52, 54, and 56 may belong to a respective one of the five taxonomies that have products that are most likely to be co-purchased with products of the taxonomy associated with the anchor product 46”; [0068] – “using aggregated co-purchase transactional data”),
yet Al Jadda/Canal/Segev does not explicitly disclose generating loss functions based on.
However, David Wei teaches a similar recommendation method (David Wei: [0005]), including
generating loss functions based on (David Wei: [0188] – “The algorithm(s) used may be configured to optimize for different loss functions”).
It would have been obvious to one of ordinary skill in the art at the time the invention was filed to have included the loss functions of David Wei in the system of Al Jadda/Canal/Segev because Al Jadda/Canal/Segev already discloses an analysis and David Wei is merely demonstrating how the analysis may occur. Additionally, it would have been obvious to have included generating loss functions based on as taught by David Wei because loss functions are well-known and the use of it in a recommender system would have helped in information prediction (David Wei: [0188]).
In regards to claim 6, Al Jadda/Canal/Segev/David Wei teaches the system of claim 5. Yet Al Jadda does not explicitly disclose wherein training the machine-learning model further comprises at least one of: optimizing the loss functions using gradient descent; or updating the item-level embedding Gaussian distributions, the user embedding, and the product-type embedding Gaussian mixture distributions based on hyperparameters of the loss functions.
However, David Wei further teaches
wherein training the machine-learning model further comprises at least one of: optimizing the loss functions using gradient descent; or updating the item-level embedding Gaussian distributions, the user embedding, and the product-type embedding Gaussian mixture distributions based on hyperparameters of the loss functions (David Wei: [0188] – “Such algorithms or models can be trained using a variety of approaches, including gradient descent, gradient boosted trees, random forests, support vector regression, or other suitable method. The algorithm(s) used may be configured to optimize for different loss functions.”).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine David Wei with Al Jadda/Canal/Segev for the reasons identified above with respect to claim 5.
The examiner notes that the limitations are recited in the alternate and, accordingly, only one need occur.
In regards to claims 15-16, all the limitations in method claims 15-16 are closely parallel to the limitations of system claims 5-6 analyzed above and rejected on the same bases.
Claims 9-10 and 19-20 are rejected under 35 U.S.C. 103 as being unpatentable over Al Jadda , in view of Canal, in view of Ghahramani et al. (US 20180114113 A1), hereinafter Ghahramani.
In regards to claim 9, Al Jadda/Canal teaches the system of claim 1. Al Jadda further discloses wherein the operations further comprise: training a product-type embedding for items for a product-type pair of the product-type pairs, wherein the training comprises: retrieving respective representations of each product type of the product-type pairs (Al Jadda: [0046] – “Candidate pairs (e.g., potential products that may be recommended as in a product collection with the anchor product) may be generated, and those candidate pairs may be ranked”; [0060-0061] – “inputting data respective of the anchor product and of each of the plurality of products in the taxonomy into a trained neural network to generate a vector respective of each product… A cosine similarity function may be applied to the respective vectors of each pairing of anchor product and second product to determine which second product(s) from the plurality of products is/are most similar to the anchor product within the taxonomy…a biLSTM may be used in various embodiments to determine similar products to an anchor product…generate text embeddings for products…The embedding outputs F(x.sub.i), F(x.sub.j) may be joined by a metric function at the top. Then, a cosine similarity…may be used” [0060] – “The neural network may have been trained on manually-created sets of similar products”); and
determining a non-personalized product-type complementarity metric at a global level between each product type of the product-type pairs using a non-personalized component (Al Jadda: [0055] – “A predefined number of closest taxonomies may be determined for the anchor product taxonomy, in some embodiments. For example, five (5) closest taxonomies may be determined based on co-purchase data”; [0003] – “determining, by the one or more processors, a second taxonomy closest to the first taxonomy. The second taxonomy may be associated with a group of products, the first taxonomy and the second taxonomy have at least a common highest taxonomy level, and the determination that the second taxonomy is closest to the first taxonomy may be made at least in part based on co-purchase data indicating that the anchor product and at least one product in the group of products are purchased together more often than products associated with other taxonomies are purchased with the anchor product”); and
determining a personalized product-type complementarity metric at an individual level between each product type of the product-type pair using a personalized component (Al Jadda: [0046] – “Candidate pairs (e.g., potential products that may be recommended as in a product collection with the anchor product) may be generated, and those candidate pairs may be ranked”; [0060-0061] – “inputting data respective of the anchor product and of each of the plurality of products in the taxonomy into a trained neural network to generate a vector respective of each product… A cosine similarity function may be applied to the respective vectors of each pairing of anchor product and second product to determine which second product(s) from the plurality of products is/are most similar to the anchor product within the taxonomy…a biLSTM may be used in various embodiments to determine similar products to an anchor product…generate text embeddings for products…The embedding outputs F(x.sub.i), F(x.sub.j) may be joined by a metric function at the top. Then, a cosine similarity…may be used”; [0059] – “for a given taxonomy from the closest taxonomies determined in step 40, each product within the taxonomy is evaluated for its similarity to the anchor product”),
yet Al Jadda does not explicitly disclose the analysis including Gaussian mixture distributions; Gaussian mixture representations; and using a probability product kernel with a weight.
However, Ghahramani teaches a similar machine learning model (Ghahramani: [abstract]), including
Gaussian mixture distributions; Gaussian mixture representations (Ghahramani: [0049] – “The expected weight distribution may follow various patterns or types, such as a Gaussian or other probabilistic distribution of the direct weights, and may be represented as a mixture model, multi-modal Gaussian”); and
using a probability product kernel with a weight (Ghahramani: [0050] – “generate the expected weight distribution θ. Thus, the indirect network 220 may include multivariate or univariate models. The indirect network 220 may be a parametric model or neural network, but may also apply to nonparametric models, such as kernel functions or Gaussian Processes…effectively characterize the expected weight distribution and have indirect parameters 280 that may be trained”).
It would have been obvious to one of ordinary skill in the art at the time the invention was filed to have included the Gaussian mixture and probability kernel of Ghahramani in the system of Al Jadda/Canal because Al Jadda/Canal already discloses an analysis and Ghahramani is merely demonstrating how the analysis may occur. Additionally, it would have been obvious to have included Gaussian mixture distributions; Gaussian mixture representations; and using a probability product kernel with a weight as taught by Ghahramani because Gaussian mixture and kernels are well-known and the use of it in a recommender system would have improved calculations (Ghahramani: [0075]).
In regards to claim 10, Al Jadda/Canal/Ghahramani teaches the system of claim 9. Al Jadda further discloses wherein training the product-type embedding further comprises: the non-personalized product-type complementarity metric and the personalized product-type complementarity metric (Al Jadda: [0055] – “A predefined number of closest taxonomies may be determined for the anchor product taxonomy, in some embodiments. For example, five (5) closest taxonomies may be determined based on co-purchase data”; [0003] – “determining, by the one or more processors, a second taxonomy closest to the first taxonomy. The second taxonomy may be associated with a group of products, the first taxonomy and the second taxonomy have at least a common highest taxonomy level, and the determination that the second taxonomy is closest to the first taxonomy may be made at least in part based on co-purchase data indicating that the anchor product and at least one product in the group of products are purchased together more often than products associated with other taxonomies are purchased with the anchor product”; [0046] – “Candidate pairs (e.g., potential products that may be recommended as in a product collection with the anchor product) may be generated, and those candidate pairs may be ranked”; [0060-0061] – “inputting data respective of the anchor product and of each of the plurality of products in the taxonomy into a trained neural network to generate a vector respective of each product… A cosine similarity function may be applied to the respective vectors of each pairing of anchor product and second product to determine which second product(s) from the plurality of products is/are most similar to the anchor product within the taxonomy…a biLSTM may be used in various embodiments to determine similar products to an anchor product…generate text embeddings for products…The embedding outputs F(x.sub.i), F(x.sub.j) may be joined by a metric function at the top. Then, a cosine similarity…may be used”; [0059] – “for a given taxonomy from the closest taxonomies determined in step 40, each product within the taxonomy is evaluated for its similarity to the anchor product),
yet Al Jadda does not explicitly disclose Gaussian mixture distributions; and generating loss functions based on.
However, Ghahramani further teaches
Gaussian mixture distributions (Ghahramani: [0049] – “The expected weight distribution may follow various patterns or types, such as a Gaussian or other probabilistic distribution of the direct weights, and may be represented as a mixture model, multi-modal Gaussian”); and
generating loss functions based on (Ghahramani: [0031] – “a loss function is used to evaluate updates to the distribution based on error to the data term generated from the prior weight distribution and error for an updated weight distribution. The loss function is used to update the expected prior distribution of direct weights and accordingly update the indirect parameters”).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine Ghahramani with Al Jadda/Canal for the reasons identified above with respect to claim 9.
In regards to claims 19-20, all the limitations in method claims 19-20 are closely parallel to the limitations of system claims 9-10 analyzed above and rejected on the same bases.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
Liu et al. (US 10861077 B1) teaches a recommendation system for increasing diversity of items. Gaussian distribution may be used in the analysis of items to determine a likelihood that items are harmonious.
NPL Reference U teaches an analysis that may be used to make dietary recommendations. Variables in analysis may be transformed into variables that have Gaussian distributions.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to ANNA MAE MITROS whose telephone number is (571)272-3969. The examiner can normally be reached Monday-Friday from 9:30-6.
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, 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.
/ANNA MAE MITROS/Examiner, Art Unit 3689