# US 12,002,077 B2 · Automatic listing generation for multiple items > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 12,002,077 B2 - Document ID: US12002077B2 - Status: Granted - Kind: B2 - Application number: 16/729,438 - Filed: 2019-12-29 - Published: 2021-07-01 - Granted: 2024-06-04 - Earliest claimed: 2019-12-29 - Assignees: eBay Inc. - Applicants: eBay Inc. - Inventors: Hoang Trinh; Huy Quang Nguyen - Web page: https://patentgrep.com/patent/US12002077B2 ## CPC - G06Q 30/0603: {Catalogue creation or management} - G06F 3/0482: Interaction with lists of selectable items, e.g. menus - G06F 16/55: Clustering; Classification - G06F 16/5866: {using information manually generated, e.g. tags, keywords, comments, manually generated location and time information} - G06F 18/22: Matching criteria, e.g. proximity measures - G06F 18/23: Clustering techniques - G06F 18/2321: using statistics or function optimisation, e.g. modelling of probability density functions - G06V 10/56: relating to colour - G06V 10/758: {Involving statistics of pixels or of feature values, e.g. histogram matching} - G06V 10/761: {Proximity, similarity or dissimilarity measures} - G06V 10/763: {Non-hierarchical techniques, e.g. based on statistics of modelling distributions} - G06V 10/764: using classification, e.g. of video objects - G06V 40/10: Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands - G06V 40/20: Movements or behaviour, e.g. gesture recognition (recognition of facial expressions G06V40/16) - G06Q: INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR - G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N) - G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING ## Abstract A machine is configured to automatically generate listings for multiple items. For example, the machine receives, from a client device, two or more images and two or more descriptions. The two or more images depict two or more items. The two or more descriptions pertain to the two or more items. The machine matches one or more images of the two or more images to a description of the two or more descriptions. The one or more images depict an item of the two or more items. The description pertains to the item. The machine, based on the matching, generates a listing of the item. The listing includes the one or more images depicting the item, and the description pertaining to the item. The machine causes display of the listing of the item in a user interface. ## Claims ### Claim 1 A method comprising: receiving, from a client device, multiple images depicting two or more items and two or more descriptions pertaining to the two or more items; extracting, using one or more machine learning models, two or more words from three or more images of the multiple images, the three or more images depicting an item of the two or more items, the two or more words describing attributes associated with the item depicted in the three or more images; selecting one or more words from the two or more words based on the one or more words being extracted from a majority of the three or more images depicting the item; generating an attribute vector by applying a word vectorization technique to the one or more words; generating two or more description vectors by applying the word vectorization technique to the two or more descriptions; matching the attribute vector to a description vector of the two or more description vectors based on comparisons between the attribute vector and the two or more description vectors, the description vector corresponding to a description pertaining to the item; generating, responsive to the matching, a listing of the item, the listing including the three or more images and the description; and causing display of the listing of the item in a user interface. ### Claim 2 (depends on 1) The method of claim 1, wherein the three or more images include a first image depicting the item and a second image depicting the item, and wherein the extracting includes: generating, using an image embedding technique, a first vector representing the first image, and a second vector representing the second image; generating a cluster of vectors by matching the first vector and the second vector; and extracting at least one word from both the first image and the second image included in the cluster of vectors, the attribute vector generated by applying the word vectorization technique to the at least one word extracted from both the first image and the second image. ### Claim 3 (depends on 2) The method of claim 2, wherein the generating the cluster of vectors includes: generating a plurality of pairs of vectors based on the multiple images depicting the two or more items; for each pair of vectors of the plurality of pairs of vectors, determining, using a distance function, a distance value between the vectors included in each pair; determining that the distance value for a pair of vectors that includes the first vector and the second vector is below a threshold value; and grouping the first vector and the second vector in the cluster of vectors based on the pair of vectors that includes the first vector and the second vector being below the threshold value. ### Claim 4 (depends on 1) The method of claim 1, wherein the matching the attribute vector to the description vector includes using a bipartite matching algorithm that performs pairwise comparisons on two or more attribute vectors associated with the two or more items and the two or more description vectors, the two or more attribute vectors including the attribute vector generated for the item. ### Claim 5 (depends on 1) The method of claim 1, wherein the multiple images depicting the two or more items are unlabeled, unordered, and included in a file. ### Claim 6 (depends on 1) The method of claim 1, wherein the two or more descriptions pertaining to the two or more items are unlabeled and included in a file. ### Claim 7 (depends on 1) The method of claim 1, wherein the two or more words include an additional word having been extracted from less than the majority of the three or more images depicting the item, the attribute vector generated by applying the word vectorization technique to the one or more words and not the additional word. ### Claim 8 (depends on 1) The method of claim 1, wherein generating the listing includes automatically adding, to the description, a word of the two or more words extracted from the three or more images, the word previously excluded from the description. ### Claim 9 A system comprising: one or more hardware processors; and a machine-readable medium for storing instructions that, when executed by the one or more hardware processors, cause the one or more hardware processors to perform operations comprising: receiving, from a client device, multiple images depicting two or more items and two or more descriptions pertaining to the two or more items; extracting, using one or more machine learning models, two or more words from three or more images of the multiple images, the three or more images depicting an item of the two or more items, the two or more words describing attributes associated with the item depicted in the three or more images; selecting one or more words from the two or more words based on the one or more words being extracted from a majority of the three or more images depicting the item; generating an attribute vector by applying a word vectorization technique to the one or more words; generating two or more description vectors by applying the word vectorization technique to the two or more descriptions; matching the attribute vector to a description vector of the two or more description vectors based on comparisons between the attribute vector and the two or more description vectors, the description vector corresponding to a description pertaining to the item; generating, responsive to the matching, a listing of the item, the listing including the three or more images and the description; and causing display of the listing of the item in a user interface. ### Claim 10 (depends on 9) The system of claim 9, wherein the three or more images include a first image depicting the item and a second image depicting the item, and wherein the extracting includes: generating, using an image embedding technique, a first vector representing the first image, and a second vector representing the second image; generating a cluster of vectors by matching the first vector and the second vector; and extracting at least one word from both the first image and the second image included in the cluster of vectors, the attribute vector generated by applying the word vectorization technique to the at least one word extracted from both the first image and the second image. ### Claim 11 (depends on 10) The system of claim 10, wherein the generating the cluster of vectors includes: generating a plurality of pairs of vectors based on the multiple images depicting the two or more items; for each pair of vectors of the plurality of pairs of vectors, determining, using a distance function, a distance value between the vectors included in each pair; determining that the distance value for a pair of vectors that includes the first vector and the second vector is below a threshold value; and grouping the first vector and the second vector in the cluster of vectors based on the pair of vectors that includes the first vector and the second vector being below the threshold value. ### Claim 12 (depends on 9) The system of claim 9, wherein the matching the attribute vector to the description vector includes using a bipartite matching algorithm that performs pairwise comparisons on two or more attribute vectors associated with the two or more items and the two or more description vectors, the two or more attribute vectors including the attribute vector generated for the item. ### Claim 13 (depends on 9) The system of claim 9, wherein the multiple images depicting the two or more items are unlabeled, unordered, and included in a first file, and the two or more descriptions pertaining to the two or more items are unlabeled and included in a second file. ### Claim 14 (depends on 9) The system of claim 9, wherein the two or more words include an additional word having been extracted from less than the majority of the three or more images depicting the item, the attribute vector generated by applying the word vectorization technique to the one or more words and not the additional word. ### Claim 15 A non-transitory machine-readable medium comprising instructions, which when executed by one or more hardware processors, cause the one or more hardware processors to perform operations comprising: receiving, from a client device, multiple images depicting two or more items and two or more descriptions pertaining to the two or more items; extracting, using one or more machine learning models, two or more words from three or more images of the multiple images, the three or more images depicting an item of the two or more items, the two or more words describing attributes associated with the item depicted in the three or more images; selecting one or more words from the two or more words based on the one or more words being extracted from a majority of the three or more images depicting the item; generating an attribute vector by applying a word vectorization technique to the one or more words; generating two or more description vectors by applying the word vectorization technique to the two or more descriptions; matching the attribute vector to a description vector of the two or more description vectors based on comparisons between the attribute vector and the two or more description vectors, the description vector corresponding to a description pertaining to the item; generating, responsive to the matching, a listing of the item, the listing including the three or more images and the description; and causing display of the listing of the item in a user interface. ### Claim 16 (depends on 15) The non-transitory machine-readable medium of claim 15, wherein the three or more images include a first image depicting the item and a second image depicting the item, and wherein the extracting includes: generating, using an image embedding technique, a first vector representing the first image, and a second vector representing the second image; generating a cluster of vectors by matching the first vector and the second vector; and extracting at least one word from both the first image and the second image included in the cluster of vectors, the attribute vector generated by applying the word vectorization technique to the at least one word extracted from both the first image and the second image. ### Claim 17 (depends on 16) The non-transitory machine-readable medium of claim 16, wherein the generating the cluster of vectors includes: generating a plurality of pairs of vectors based on the multiple images depicting the two or more items; for each pair of vectors of the plurality of pairs of vectors, determining, using a distance function, a distance value between the vectors included in each pair; determining that the distance value for a pair of vectors that includes the first vector and the second vector is below a threshold value; and grouping the first vector and the second vector in the cluster of vectors based on the pair of vectors that includes the first vector and the second vector being below the threshold value. ### Claim 18 (depends on 15) The non-transitory machine-readable medium of claim 15, wherein the matching the attribute vector to the description vector includes using a bipartite matching algorithm that performs pairwise comparisons on two or more attribute vectors associated with the two or more items and the two or more description vectors, the two or more attribute vectors including the attribute vector generated for the item. ### Claim 19 (depends on 15) The non-transitory machine-readable medium of claim 15, wherein the multiple images depicting the two or more items are unlabeled, unordered, and included in a first file, and wherein the two or more descriptions pertaining to the two or more items are unlabeled and included in a second file. ### Claim 20 (depends on 15) The non-transitory machine-readable medium of claim 15, wherein the two or more words include an additional word having been extracted from less than the majority of the three or more images depicting the item, the attribute vector generated by applying the word vectorization technique to the one or more words and not the additional word.