-
公开(公告)号：US11676015B2
公开(公告)日：2023-06-13
申请号：US16886470
申请日：2020-05-28
Applicant: salesforce.com, inc.
Inventor： Arthur Zhang , Joshua Correa
IPC: G06N3/08 , G06N20/20 , G06Q30/0601 , G06N7/01
CPC classification number: G06N3/08 , G06N7/01 , G06N20/20 , G06Q30/0631
Abstract: Systems, devices, and techniques are disclosed for recommendations using a transformer neural network. User activity data including items and actions associated with users and a catalog including descriptions of the items may be received. User vectors for the users, item vectors for the items and action vectors the actions may be generated by applying singular vector decomposition to the user activity data. Sequence vectors may be generated based on item vectors and the action vectors. Transformer vectors may be generated by applying a text-to-text transferring transformer to descriptions of the items. Similarity vectors may be generated based on the transformer vectors. Merged vectors may be generated by merging the sequence vector, transformer vector, and similarity vector for items. A set of probabilities may be determined by inputting the user vector for the user, merged vectors for the items, and sequence vectors for the actions to a transformer neural network.
-
公开(公告)号：US20220058713A1
公开(公告)日：2022-02-24
申请号：US16999845
申请日：2020-08-21
Applicant: salesforce.com, inc.
Inventor： Joshua Correa , Alexander Kushkuley
Abstract: A group of recommendations related to an item, such as an item of content presented to a user in a page, can be ranked according to a probability distribution that is iteratively updated with each user interaction. For practical implementations, a click stream of interactions may be logged, and then applied in a batch process to update the probability distribution on any suitable schedule independent of the timing of incoming user interactions.
-
Patent Agency Ranking
公开(公告)号：US11276104B1
公开(公告)日：2022-03-15
申请号：US16999845
申请日：2020-08-21
Applicant: salesforce.com, inc.
Inventor： Joshua Correa , Alexander Kushkuley
Abstract: A group of recommendations related to an item, such as an item of content presented to a user in a page, can be ranked according to a probability distribution that is iteratively updated with each user interaction. For practical implementations, a click stream of interactions may be logged, and then applied in a batch process to update the probability distribution on any suitable schedule independent of the timing of incoming user interactions.
-
公开(公告)号：US10762548B1
公开(公告)日：2020-09-01
申请号：US15181594
申请日：2016-06-14
Applicant: salesforce.com, inc.
Inventor： Bharath K. Krishnan , Rene Borm , Joshua Correa , Rene Kessler , Peter Koch , Vishwamitra S. Ramakrishnan
Abstract: The foregoing are among the objects attained by the invention, which provides, in some aspects, digital data processing methods for generation of customized user interfaces that present links, images or other components representing items of interest to a user in an order that is prioritized as a function of (a) representations in a multidimensional factor space of characteristics of the respective items, and (b) representations in that same multidimensional space of characteristics of “context-indicative items”—e.g., items in which the user has previously shown an interest, as indicated by clicks or other interactions with those items respective components in the user interface.
-
公开(公告)号：US20230245206A1
公开(公告)日：2023-08-03
申请号：US17589657
申请日：2022-01-31
Applicant: salesforce.com, inc.
Inventor： Joshua Correa , Alexander Kushkuley
CPC classification number: G06Q30/0631 , G06Q30/0201 , G06F11/3438 , G06F11/3476
Abstract: A method and system for item-to-item recommendation that collects a set of visitors having interacted with at least one product of a website containing a collection of products, creates a click matrix including a collection of per-product visitor sets based on the set of visitors, change a weight value for at least one of the set of visitors, construct a co-view matrix based on determining a product of each of the changed set of visitors for each pair of products of the collection of products, determine a per-product ordered ranking of product pairs based on the co-view matrix, and select a recommended product based on a user selected product and the per-product ordered ranking of product pairs.
-
公开(公告)号：US20220188900A1
公开(公告)日：2022-06-16
申请号：US17688159
申请日：2022-03-07
Applicant: salesforce.com, inc.
Inventor： Joshua Correa , Alexander Kushkuley
Abstract: A group of recommendations related to an item, such as an item of content presented to a user in a page, can be ranked according to a probability distribution that is iteratively updated with each user interaction. For practical implementations, a click stream of interactions may be logged, and then applied in a batch process to update the probability distribution on any suitable schedule independent of the timing of incoming user interactions.
-
公开(公告)号：US20210374520A1
公开(公告)日：2021-12-02
申请号：US16886470
申请日：2020-05-28
Applicant: salesforce.com, inc.
Inventor： Arthur Zhang , Joshua Correa
Abstract: Systems, devices, and techniques are disclosed for recommendations using a transformer neural network. User activity data including items and actions associated with users and a catalog including descriptions of the items may be received. User vectors for the users, item vectors for the items and action vectors the actions may be generated by applying singular vector decomposition to the user activity data. Sequence vectors may be generated based on item vectors and the action vectors. Transformer vectors may be generated by applying a text-to-text transferring transformer to descriptions of the items. Similarity vectors may be generated based on the transformer vectors. Merged vectors may be generated by merging the sequence vector, transformer vector, and similarity vector for items. A set of probabilities may be determined by inputting the user vector for the user, merged vectors for the items, and sequence vectors for the actions to a transformer neural network.
-
-
-
-
-
-
Search Results
7
records
(took 0.017 seconds)
