DETAILED ACTION
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 .
Status of Claims
Applicant's “Amendment” filed on 8/10/2026 has been considered.
Rejection to Claims 1-20 under 35 USC 101 have not been overcome.
Claims 1-2, 5-6, 8-11, 17-18 are amended.
Claims 1-20 are currently pending and have been examined.
Claim Rejections - 35 USC § 103
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 of this title, 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.
Claims 1-20 are rejected under 35 U.S.C. 103 as being unpatentable over US 20230214432 A1 to TUNKELANG in view of US 11315164 B2 to BHATTACHARJEE in further view of US 2024/0303711 A1 to Tan.
Regarding Claim 1, TUNKELANG discloses One or more computer storage media storing computer-useable instructions that, when used by one or more computing devices of an listing platform, cause the one or more computing devices to perform operations, the operations comprising:
receiving a textual indication of a seed item listing; ([Claim 8] receiving a search query; identifying a related collected search query from collected search queries within an item dataset, the item dataset comprising the collected search queries associated with aggregate item description vectors for items listed on an item database, wherein the related collected search query is identified based on a search query vector and an aggregate item description vector for the related collected search query; and providing search results determined for the related collected search query in response to receiving the search query. [0049] To identify and return search results, client device 104 may receive a search query at a search engine. A search query can be provided in any form, including text, images, or the like. )
and providing an indication of the one or more item listings for presentation. ([0029] Using different learning techniques, such as nearest neighbor analyses and clustering algorithms, related collected search queries can be identified, and the search results associated with these identified related collected search queries can be provided.)
But does not explicitly disclose corresponding to an item listing from the listing platform; for the listing platform; receiving a textual indication of a seed item listing corresponding to an item listing from the listing platform; responsive to receiving the textual indication of the seed item listing and without user input: programmatically generating, from the textual indication of the seed item listing, a prompt for a generative model; providing the prompt to the generative model to generate a plurality of synthetic items, each synthetic item comprising text describing an item that is similar to or complementary to the seed item listing; for each of the plurality of synthetic items, performing a search on an item listings data store for the listing platform using a query generated from the text for the synthetic item to identify one or more item listings from the item listings data store, wherein generating the plurality of synthetic items and identifying the one or more item listings for each of the plurality of synthetic items reduces a number of user-submitted search queries processed by the listing platform to identify relevant item listings; and generating a set of item listings based on relevance of each of the one or more item listings identified for each of the plurality of synthetic items; and providing, for presentation on a user device, a user interface of the listing platform presenting an indication of the one or more set of item listings for presentation..
BHATTACHARJEE, on the other hand, teaches corresponding to an item listing from the listing platform; for the listing platform ([Col 2 Ln 45-50] an online retail site. [Col 4 Ln 60-65] a listing of a top 3-10 or more items based on complementarity scores generated by the recommendation modeling engine 112. [Col 5 Ln 25-30] a listing of items known to be complementary to each other,)
It would have been obvious to one of ordinary skill in the art at the time the invention was filed to modify the system of TUNKELANG to include the features, as taught by BHATTACHARJEE. One of ordinary skill in the art at the time of the invention was filed would have been motivated to modify TUNKELANG in order to provide items of interest to the customer (BHATTACHARJEE, [Col 1 Ln 20-25]).
Tan, on the other hand, teaches receiving a textual indication of a seed item listing corresponding to an item listing from the listing platform; ([0054] The system receives 300 a natural language query from a client device of a user. )
responsive to receiving the textual indication of the seed item listing and without user input: programmatically generating, from the textual indication of the seed item listing, a prompt for a generative model; ([0060] The system generates 320 a prompt for input to a machine learning based language model based on the natural language query and the contextual information.)
providing the prompt to the generative model to generate a plurality of synthetic items, each synthetic item comprising text describing an item that is similar to or complementary to the seed item listing; ([0061] The system provides 330 the prompt to the machine learning based language model for execution. The system receives 340 a response to the prompt from the machine learning based language model. The response may be a parse-able string of text that the system may further analyze to identify individual attributes. The response may include a set of attributes including, for example, specific details of the types of items that the user may be interested in. For example, the response may identify the type of food and drinks that may be relevant to the party.)
for each of the plurality of synthetic items, performing a search on an item listings data store for the listing platform using a query generated from the text for the synthetic item to identify one or more item listings from the item listings data store, wherein generating the plurality of synthetic items and identifying the one or more item listings for each of the plurality of synthetic items reduces a number of user-submitted search queries processed by the listing platform to identify relevant item listings; and ([0063] The system further generates 350 a search query based on the set of attributes. The search query may be a query for searching for items in a database, for example, a product catalog of retail stores.)
generating a set of item listings based on relevance of each of the one or more item listings identified for each of the plurality of synthetic items; and ([Col 2 Ln 45-50] an online retail site. [Col 4 Ln 60-65] a listing of a top 3-10 or more items based on complementarity scores generated by the recommendation modeling engine 112. [Col 5 Ln 25-30] a listing of items known to be complementary to each other,)
providing, for presentation on a user device, a user interface of the listing platform presenting an indication of the one or more set of item listings for presentation.. ([0064] The system receives search results in response to the search query and provides 370 the search results to the client device of the user. The system may send a message to the user with the search results. The system may add one or more items selected based on the search to a shopping cart of the user. )
It would have been obvious to one of ordinary skill in the art at the time the invention was filed to modify the system of TUNKELANG to include the features, as taught by Tan. One of ordinary skill in the art at the time of the invention was filed would have been motivated to modify TUNKELANG in order to enhance search queries (Tan, [0002]).
Regarding Claim 2, TUNKELANG in view of BHATTACHARJEE and Tan teaches the one or more computer storage media of claim 1.
TUNKELANG discloses wherein programmatically generating the prompt and providing the prompt to the generative model comprises: in response to receiving the textual indication of the seed item listing, generating a prompt based on the seed item listing; and providing the prompt as input to the generative model, causing the generative model to generate the plurality of synthetic items. ([0005-0006] The search results with which users have interacted can be represented in the vector space as item description vectors, which comprise vector representations of a portion of each search result, such as a title, image, description, or the like. As a result, based on the interaction tracking, each of the collected search queries is associated with one or more item description vectors corresponding to the search results with which users have interacted. The item description vectors for each respective collected search query can be aggregated. This can include averaging the item description vectors. Aggregating the item description vectors generates aggregate item description vectors, which are each respectively associated with the collected search queries.)
Regarding Claim 3, TUNKELANG in view of BHATTACHARJEE and Tan teaches the one or more computer storage media of claim 2.
TUNKELANG discloses wherein the prompt includes an item title for the seed item listing. ([0005-0006] The search results with which users have interacted can be represented in the vector space as item description vectors, which comprise vector representations of a portion of each search result, such as a title, image, description, or the like. As a result, based on the interaction tracking, each of the collected search queries is associated with one or more item description vectors corresponding to the search results with which users have interacted. The item description vectors for each respective collected search query can be aggregated. This can include averaging the item description vectors. Aggregating the item description vectors generates aggregate item description vectors, which are each respectively associated with the collected search queries.)
Regarding Claim 4, TUNKELANG in view of BHATTACHARJEE and Tan teaches the one or more computer storage media of claim 2.
TUNKELANG discloses wherein the prompt is generated using a predefined prompt template selected based on a category of the seed item.. ([0071] Classification prediction is further beneficial when using search engines for websites or other databases that categorize data. In the context of an online marketplace, many items, which can be provided as search results, are associated with particular categories, and thus, predicting a category for a search query and using it as an input to trained neural network 132 is beneficial in identifying related search queries, and ultimately inferring the search query intent. [0072] Classifier engine 114 can be applied to a received search query to predict the classification, e.g., a category, for the search query. In predicting a classification, classifier engine 114 can employ trained classifier 134, where classifier engine 114 provides a search query input to trained classifier 134, which in response, outputs a predicted classification for the search query input. .)
Regarding Claim 5, TUNKELANG in view of BHATTACHARJEE and Tan teaches the one or more computer storage media of claim 1.
TUNKELANG discloses wherein generating the plurality of synthetic items comprises obtaining a textual indication of the plurality of synthetic items generated by the generative model, wherein obtaining comprises: performing a lookup on a key-value store based on the textual indication of the seed item listing to identify a matching item listing, the key-value data store storing an indication of each of a plurality of item listings for which the generative model has generated synthetic items; and retrieving the plurality of synthetic items stored in associated with the matching item listing. ([0007] At runtime, a search query is received. The search query can be matched to a collected search query. A search query vector can then be identified for the search query by using the aggregate item description vector for the matching collected search query. In another case, the search query vector is predicted, e.g., “inferred,” using a neural network having been trained on aggregate item description vectors and associated collected search queries. [0008] search results associated with the related collected search queries are returned for the received search query. As noted, the identified related collected search queries may better represent the intent of a user that is associated with the received search query.)
Regarding Claim 6, TUNKELANG in view of BHATTACHARJEE and Tan teaches the one or more computer storage media of claim 1.
TUNKELANG discloses wherein generating the plurality of synthetic items comprises obtaining a textual indication of the plurality of synthetic items generated by the generative model, wherein obtaining comprises: performing a search on an approximate match data store based on the textual indication of the seed item listing to identify a similar item listing, the approximate match data storing an indication of each item listing for which the generative model has generated synthetic items; and retrieving the plurality of synthetic items stored for the similar item listing. ([0019] Sometimes query similarity is evident from surface features of the queries. These surface variations can include: (1) Word inflection. The most common case of variation in word inflection is using the singular and plural forms of nouns (e.g., birkenstock=birkenstocks). Queries can be canonicalized by applying stemming or lemmatization to queries. (2) Word order. Queries may use the same words in a different order (e.g., earbuds samsung=samsung earbuds). Queries can be canonicalized by sorting the query tokens in alphabetical (lexicographic) order. (3) Compounding. Queries may vary in compounding or decompounding query tokens (e.g., backpack=back pack). Queries can be canonicalized by removing spaces and other token separators. [0008] The search query vector can be compared to aggregate item description vectors in the dataset. For instance, a distance measurement can be used for the comparison. Other methods for performing the comparison include clustering and nearest neighbor algorithms. Based on the comparison, aggregate item description vectors are identified for the search query vector. The identified aggregate item description vectors are associated with collected search queries that are related to the received search query. These related collected search queries can be returned as search query recommendations. In another case, search results associated with the related collected search queries are returned for the received search query. As noted, the identified related collected search queries may better represent the intent of a user that is associated with the received search query.)
Regarding Claim 7, TUNKELANG in view of BHATTACHARJEE and Tan teaches the one or more computer storage media of claim 6.
BHATTACHARJEE teaches wherein performing the search on the approximate search data store comprises: generating a seed item embedding from the textual indication of the seed item listing; and determining a similarity between the seed item embedding and an item embedding for each of one or more item listings in the approximate match data store. ([Col 6 Ln 1-20] Generating a complementary item model involves performing a pairwise vectorization of item descriptions to generate embeddings indicative of similarity between two item descriptions, and then classifying those vectorized descriptions to determine the extent of complementarity. As noted above, the vectorizations of the items that are indicated as complementary can be performed using embeddings generated using Mikolov et. al's Doc2Vec/Paragraph2Vec model. Subsequent scoring of item pairs can be performed, for example, using a Siamese network as discussed below in connection with FIGS. 4-5. The Siamese network may be trained, for example, using known complementary items (using 1 as the value for known complementarity), and optionally a random assortment of known non-complementary items (using 0 as the value for known non-complementarity).)
It would have been obvious to one of ordinary skill in the art at the time the invention was filed to modify the system of TUNKELANG to include the features, as taught by BHATTACHARJEE. One of ordinary skill in the art at the time of the invention was filed would have been motivated to modify TUNKELANG in order to provide items of interest to the customer (BHATTACHARJEE, [Col 1 Ln 20-25]).
Regarding Claim 8, TUNKELANG in view of BHATTACHARJEE and Tan teaches the one or more computer storage media of claim 6.
TUNKELANG discloses wherein obtaining further comprises: performing a lookup on a key-value store based on the textual indication of the seed item listing; and determining an exact match for the seed item listing is absent from the key-value store, wherein the search on the approximate match data store is performed in response to determining an exact match for the seed item listing is absent from the key-value store. ([0037] A neural network can be trained on the item dataset so that the trained neural network is configured to receive a search query input and output a predicted, i.e., “inferred,” search query vector. This is particularly beneficial in situations where a search query is received and there is no matching collected search query. For instance, the neural network can be employed to identify a search query vector for the search query, and the search query vector can be used to identify a related collected search query from among the collected search queries of the item dataset by comparing the search query vector to the aggregate item description vectors, for example, using a distance measurement, nearest neighbor algorithm, clustering method, or the like. The related collected search query can be similarly treated in that it can be provided as a recommendation or the search results associated with the related search query can be identified and provided in response to the received search query.)
Regarding Claim 9, TUNKELANG in view of BHATTACHARJEE and Tan teaches the one or more computer storage media of claim 1.
TUNKELANG discloses wherein generating the plurality of synthetic items comprises obtaining a textual indication of the plurality of synthetic items generated by the generative model, wherein obtaining comprises: performing a search on an approximate match data store based on the textual indication of the seed item listing, the approximate match data store storing an indication of item listing for which the generative model has generated synthetic items; and in response to determining there is no similar item in the approximate match data store having a similarity score that satisfies a similarity threshold, providing an input based on the seed item listing to the generative model, causing the generative model to generate the plurality of synthetic items. ([0037] A neural network can be trained on the item dataset so that the trained neural network is configured to receive a search query input and output a predicted, i.e., “inferred,” search query vector. This is particularly beneficial in situations where a search query is received and there is no matching collected search query. For instance, the neural network can be employed to identify a search query vector for the search query, and the search query vector can be used to identify a related collected search query from among the collected search queries of the item dataset by comparing the search query vector to the aggregate item description vectors, for example, using a distance measurement, nearest neighbor algorithm, clustering method, or the like. The related collected search query can be similarly treated in that it can be provided as a recommendation or the search results associated with the related search query can be identified and provided in response to the received search query. [0079] In some instances, a threshold distance can be used to identify the aggregated item description vectors within the threshold. In an example, the threshold distance is a cosine value, and can be set at any value, as will be understood by those of ordinary skill in the art. In some cases, the threshold distance value is determined by the variance of item vectors aggregated to form the aggregated item description vectors. That is, a cosine similarity metric can be computed between each of the item vectors aggregated to form an aggregate item description vector. [0080] The threshold distance value can then be determined based on the value of the variance. In an aspect, as the variance increases, the threshold distance value can decrease (e.g., lowering a cosine similarity from 0.98 to 0.96) to encompass a greater distance for identifying aggregate item description vectors. This can be beneficial because the higher variance might indicate a higher degree of uncertainty in the relationship of the item vectors aggregated into the aggregate item description vectors. As such, the lower threshold value, with respect to a lower cosine similarity value, provides aggregate item description vectors at a relatively greater distance to be identified within item dataset )
Claim 10 recites a method comprising substantially similar limitations as claim 1. The claim is rejected under substantially similar grounds as claim 1.
Claim 11 recites a method comprising substantially similar limitations as claim 2. The claim is rejected under substantially similar grounds as claim 2.
Claim 12 recites a method comprising substantially similar limitations as claim 3. The claim is rejected under substantially similar grounds as claim 3.
Claim 13 recites a method comprising substantially similar limitations as claim 4. The claim is rejected under substantially similar grounds as claim 4.
Regarding Claim 14, TUNKELANG in view of BHATTACHARJEE and Tan teaches the method of claim 10.
TUNKELANG discloses wherein the method further comprises: performing a search on an approximate match data store based on the textual indication of the seed item listing , the approximate match data store storing an indication of each item listing for which the generative model has generated synthetic items; and determining a similar item listing for the seed item listing is absent from the approximate match data store. ([0019] Sometimes query similarity is evident from surface features of the queries. These surface variations can include: (1) Word inflection. The most common case of variation in word inflection is using the singular and plural forms of nouns (e.g., birkenstock=birkenstocks). Queries can be canonicalized by applying stemming or lemmatization to queries. (2) Word order. Queries may use the same words in a different order (e.g., earbuds samsung=samsung earbuds). Queries can be canonicalized by sorting the query tokens in alphabetical (lexicographic) order. (3) Compounding. Queries may vary in compounding or decompounding query tokens (e.g., backpack=back pack). Queries can be canonicalized by removing spaces and other token separators. [0008] The search query vector can be compared to aggregate item description vectors in the dataset. For instance, a distance measurement can be used for the comparison. Other methods for performing the comparison include clustering and nearest neighbor algorithms. Based on the comparison, aggregate item description vectors are identified for the search query vector. The identified aggregate item description vectors are associated with collected search queries that are related to the received search query. These related collected search queries can be returned as search query recommendations. In another case, search results associated with the related collected search queries are returned for the received search query. As noted, the identified related collected search queries may better represent the intent of a user that is associated with the received search query.)
Claim 15 recites a method comprising substantially similar limitations as claim 7. The claim is rejected under substantially similar grounds as claim 7.
Claim 16 recites a method comprising substantially similar limitations as claim 8. The claim is rejected under substantially similar grounds as claim 8.
Regarding Claim 17, TUNKELANG discloses A computer system comprising: one or more processors; and one or more computer storage media storing computer-useable instructions that, when used by the one or more processors, causes the computer system to perform operations comprising:
receiving a textual indication of a seed item listing; ([Claim 8] receiving a search query; identifying a related collected search query from collected search queries within an item dataset, the item dataset comprising the collected search queries associated with aggregate item description vectors for items listed on an item database, wherein the related collected search query is identified based on a search query vector and an aggregate item description vector for the related collected search query; and providing search results determined for the related collected search query in response to receiving the search query. [0049] To identify and return search results, client device 104 may receive a search query at a search engine. A search query can be provided in any form, including text, images, or the like. )
obtaining, based on the seed item listing, a textual indication of one or more synthetic items generated by a generative model by: ([0052] collected search queries may broadly include any received or generated search query, including those received at (or from) search engines, or search queries generated using potential search results. (For instance, an item title of an item listing on an item listing database can be used to generate a search query (textual indication of one or more synthetic items) for inclusion in collected search queries. [0068] In general, vector prediction engine 112 generates a search query vector for a received search query input using trained neural network 132. Model trainer 126 of vector prediction engine 112 can be used to train a machine learning model, such as a neural network, to predict a search query vector for a search query input. Several types of neural networks may be suitable for use with the present technology. For instance, the neural network may be a perceptron or MLP (multilayer perceptron), an ANN (artificial neural network), a radial network, a CNN (convolutional neural network), an RNN (recurrent neural network), an LSTM (long short-term memory) neural network, or the like. One particular example that has been identified as suitable in a method reduced to practice is BERT (Bidirectional Encoder Representations from Transformers).))
performing a lookup on a key-value store based on the textual indication of the seed item listing, ([0059] One method of vector aggregation that can be employed by vector aggregator 120 includes averaging. That is, the item vectors comprise values that numerically represent a position in a multidimensional space relative to other item vectors. These values can be averaged to provide an averaged aggregate item description vector that comprises values representative of the average of the item vectors.) (performing an averaging of item vector values is interpreted as a key value lookup.)
when determining an exact match for the seed item listing is present in the key-value store, retrieving a textual indication of one or more synthetic items generated by the generative model for the exact match, ([0066] One method of determining the search query vector is to identify a matching collected search query having an associated aggregate item description vector representation, and then employing the associated aggregate item description vector as the search query vector. For instance, a search query engine can use item dataset 130 to determine a search query vector. Here, when the search query is received, the received search query is compared to collected search queries of item dataset 130 to determine if a collected search query matches the received search query. A matching collected search query has with it an associated aggregate item description vector. The aggregate item description vector for the matching collected search query can be used as the search query vector for the received search query.)
when determining an exact match for the seed item listing is absent in the key-value store, performing a search on an approximate match data store based on the textual indication of the seed item listing to identify a similar item listing and retrieving a textual indication of one or more synthetic items generated by the generative model for the similar item listing, the approximate match data store storing an indication of each item listing for which the generative model has generated synthetic items; and ([0018] There is also value in recognizing when two queries express similar but not necessarily equivalent intent. The search engine can leverage this similarity to increase recall without unduly decreasing precision. [0019] Sometimes query similarity is evident from surface features of the queries. These surface variations can include: (1) Word inflection. The most common case of variation in word inflection is using the singular and plural forms of nouns (e.g., birkenstock=birkenstocks). Queries can be canonicalized by applying stemming or lemmatization to queries. (2) Word order. Queries may use the same words in a different order (e.g., earbuds samsung=samsung earbuds). )
performing a search on an item listings data store based on the textual indication of the one or more synthetic items to identify one or more item listings; and providing an indication of the one or more item listings for presentation. ([0029] Using different learning techniques, such as nearest neighbor analyses and clustering algorithms, related collected search queries can be identified, and the search results associated with these identified related collected search queries can be provided.)
But does not explicitly disclose corresponding to an item listing from the listing platform; for the listing platform; receiving a textual indication of a seed item listing corresponding to an item listing from the listing platform; responsive to receiving the textual indication of the seed item listing and without user input: programmatically generating, from the textual indication of the seed item listing, a prompt for a generative model; providing the prompt to the generative model to generate a plurality of synthetic items, each synthetic item comprising text describing an item that is similar to or complementary to the seed item listing; maintaining a textual indication of the plurality of synthetic items generated by the generative model, wherein maintaining comprises: for each of the plurality of synthetic items, performing a search on an item listings data store for the listing platform using a query generated from the text for the synthetic item to identify one or more item listings from the item listings data store, wherein generating the plurality of synthetic items and identifying the one or more item listings for each of the plurality of synthetic items reduces a number of user-submitted search queries processed by the listing platform to identify relevant item listings; and generating a set of item listings based on relevance of each of the one or more item listings identified for each of the plurality of synthetic items; and providing, for presentation on a user device, a user interface of the listing platform presenting an indication of the one or more set of item listings for presentation..
BHATTACHARJEE, on the other hand, teaches corresponding to an item listing from the listing platform; for the listing platform; ([Col 2 Ln 45-50] an online retail site. [Col 4 Ln 60-65] a listing of a top 3-10 or more items based on complementarity scores generated by the recommendation modeling engine 112. [Col 5 Ln 25-30] a listing of items known to be complementary to each other,)
It would have been obvious to one of ordinary skill in the art at the time the invention was filed to modify the system of TUNKELANG to include the features, as taught by BHATTACHARJEE. One of ordinary skill in the art at the time of the invention was filed would have been motivated to modify TUNKELANG in order to provide items of interest to the customer (BHATTACHARJEE, [Col 1 Ln 20-25]).
Tan, on the other hand, teaches receiving a textual indication of a seed item listing corresponding to an item listing from the listing platform; ([0054] The system receives 300 a natural language query from a client device of a user. )
responsive to receiving the textual indication of the seed item listing and without user input: programmatically generating, from the textual indication of the seed item listing, a prompt for a generative model; ([0060] The system generates 320 a prompt for input to a machine learning based language model based on the natural language query and the contextual information.)
providing the prompt to the generative model to generate a plurality of synthetic items, each synthetic item comprising text describing an item that is similar to or complementary to the seed item listing; ([0061] The system provides 330 the prompt to the machine learning based language model for execution. The system receives 340 a response to the prompt from the machine learning based language model. The response may be a parse-able string of text that the system may further analyze to identify individual attributes. The response may include a set of attributes including, for example, specific details of the types of items that the user may be interested in. For example, the response may identify the type of food and drinks that may be relevant to the party.)
maintaining a textual indication of the plurality of synthetic items generated by the generative model, wherein maintaining comprises: for each of the plurality of synthetic items, performing a search on an item listings data store for the listing platform using a query generated from the text for the synthetic item to identify one or more item listings from the item listings data store, wherein generating the plurality of synthetic items and identifying the one or more item listings for each of the plurality of synthetic items reduces a number of user-submitted search queries processed by the listing platform to identify relevant item listings; and ([0063] The system further generates 350 a search query based on the set of attributes. The search query may be a query for searching for items in a database, for example, a product catalog of retail stores.)
generating a set of item listings based on relevance of each of the one or more item listings identified for each of the plurality of synthetic items; and ([Col 2 Ln 45-50] an online retail site. [Col 4 Ln 60-65] a listing of a top 3-10 or more items based on complementarity scores generated by the recommendation modeling engine 112. [Col 5 Ln 25-30] a listing of items known to be complementary to each other,)
providing, for presentation on a user device, a user interface of the listing platform presenting an indication of the one or more set of item listings for presentation.. ([0064] The system receives search results in response to the search query and provides 370 the search results to the client device of the user. The system may send a message to the user with the search results. The system may add one or more items selected based on the search to a shopping cart of the user. )
It would have been obvious to one of ordinary skill in the art at the time the invention was filed to modify the system of TUNKELANG to include the features, as taught by Tan. One of ordinary skill in the art at the time of the invention was filed would have been motivated to modify TUNKELANG in order to enhance search queries (Tan, [0002]).
Regarding Claim 18, TUNKELANG in view of BHATTACHARJEE and Tan teaches the computer system of claim 17.
TUNKELANG discloses wherein the key-value store includes an item listing identifier as a key for each of a plurality of item listings for which the generative model has generated the plurality of synthetic items, and wherein performing the lookup on the key-value store comprises determining whether an item listing identifier in the key-value store matches an item listing identifier for the seed item listing. ([0019] Sometimes query similarity is evident from surface features of the queries. These surface variations can include: (1) Word inflection. The most common case of variation in word inflection is using the singular and plural forms of nouns (e.g., birkenstock=birkenstocks). Queries can be canonicalized by applying stemming or lemmatization to queries. (2) Word order. Queries may use the same words in a different order (e.g., earbuds samsung=samsung earbuds). Queries can be canonicalized by sorting the query tokens in alphabetical (lexicographic) order. (3) Compounding. Queries may vary in compounding or decompounding query tokens (e.g., backpack=back pack). Queries can be canonicalized by removing spaces and other token separators. [0008] The search query vector can be compared to aggregate item description vectors in the dataset. For instance, a distance measurement can be used for the comparison. Other methods for performing the comparison include clustering and nearest neighbor algorithms. Based on the comparison, aggregate item description vectors are identified for the search query vector. The identified aggregate item description vectors are associated with collected search queries that are related to the received search query. These related collected search queries can be returned as search query recommendations. In another case, search results associated with the related collected search queries are returned for the received search query. As noted, the identified related collected search queries may better represent the intent of a user that is associated with the received search query.)
Claim 19 recites a system comprising substantially similar limitations as claim 7. The claim is rejected under substantially similar grounds as claim 7.
Regarding Claim 20, TUNKELANG in view of BHATTACHARJEE and Tan teaches the computer system of claim 19.
BHATTACHARJEE teaches wherein the seed item embedding is generated by providing an item title and/or item description of the seed item listing to an embedding model. ([Col 6 Ln 1-20] Generating a complementary item model involves performing a pairwise vectorization of item descriptions to generate embeddings indicative of similarity between two item descriptions, and then classifying those vectorized descriptions to determine the extent of complementarity. As noted above, the vectorizations of the items that are indicated as complementary can be performed using embeddings generated using Mikolov et. al's Doc2Vec/Paragraph2Vec model. Subsequent scoring of item pairs can be performed, for example, using a Siamese network as discussed below in connection with FIGS. 4-5. The Siamese network may be trained, for example, using known complementary items (using 1 as the value for known complementarity), and optionally a random assortment of known non-complementary items (using 0 as the value for known non-complementarity [Col 1 Ln 45-60] generating a model of item complementarity based at least in part on the one or more pairs of complementary items and item descriptions of each of the items included in the item collection. The method includes receiving a selection of a seed item from within the item collection, and, in response to selection of the seed item, identifying one or more complementary items. At least one of the seed item or the one or more complementary items is not included in the one or more pairs of complementary items included in the received data.).)
It would have been obvious to one of ordinary skill in the art at the time the invention was filed to modify the system of TUNKELANG to include the features, as taught by BHATTACHARJEE. One of ordinary skill in the art at the time of the invention was filed would have been motivated to modify TUNKELANG in order to provide items of interest to the customer (BHATTACHARJEE, [Col 1 Ln 20-25]).
Response to Arguments
Applicant’s arguments filed with respect to the rejection of claims under 35 USC 101 have been fully considered.
The amendments to the claims provide a practical application of the abstract idea. The claims are integrated into a practical application and the rejections under 35 USC 101 are overcome.
Applicant’s arguments with respect to rejection of the claim under 35 USC 103 have been considered but are moot in view of new grounds of rejection, necessitated by Applicant’s amendment.
Applicant argues that the cited references do not teach programmatically generating a prompt from a seed item listing, providing that prompt to a generative model to generate a plurality of synthetic items each comprising text describing an item similar to or complementary to the seed, and using each synthetic item as the basis for a separate search against the item listings data store.
However, Examiner relies on a new combination of references to teach these limitations in the claims. Examiner directs Applicant’s attention to the office action, above.
Conclusion
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to Michelle T. Kringen whose telephone number is (571)270-0159. The examiner can normally be reached M-F: 11am-7pm.
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.
/MICHELLE T KRINGEN/Primary Examiner, Art Unit 3689