公开(公告)号：US20210224284A1

公开(公告)日：2021-07-22

申请号：US16749540

申请日：2020-01-22

Applicant: salesforce.com, inc.

Inventor： Guillaume Kempf ,  Jacob Alexander Mannix ,  Arvind Srikantan

IPC: G06F16/2457 ,  G06F16/22 ,  G06F16/2452 ,  G06F16/28

Abstract: Methods, systems, and devices supporting determining user and data record relationships based on vector space embeddings are described. Some database systems may receive data record access indications corresponding to data records accessed by users. A database system may generate, based on the data record access indications, user sessions for the users, data record sessions for the data records, or a combination for users and data records. For example, a user session may correspond to a respective user and include a record identifier associated with each data record accessed by the user. The system may generate, in a vector space, vectors from the sessions using an embedding operation, where each vector corresponds to a respective user or data record. The system may determine relationships between the users, data records, or both based on the vectors and may transmit an indication of at least one data record based on the relationships.

公开(公告)号：US11475018B2

公开(公告)日：2022-10-18

申请号：US16749540

申请日：2020-01-22

Applicant: salesforce.com, inc.

Inventor： Guillaume Kempf ,  Jacob Alexander Mannix ,  Arvind Srikantan

IPC: G06F16/2457 ,  G06F16/22 ,  G06F16/2452 ,  G06F16/28

Abstract: Methods, systems, and devices supporting determining user and data record relationships based on vector space embeddings are described. Some database systems may receive data record access indications corresponding to data records accessed by users. A database system may generate, based on the data record access indications, user sessions for the users, data record sessions for the data records, or a combination for users and data records. For example, a user session may correspond to a respective user and include a record identifier associated with each data record accessed by the user. The system may generate, in a vector space, vectors from the sessions using an embedding operation, where each vector corresponds to a respective user or data record. The system may determine relationships between the users, data records, or both based on the vectors and may transmit an indication of at least one data record based on the relationships.

公开(公告)号：US20220229843A1

公开(公告)日：2022-07-21

申请号：US17154378

申请日：2021-01-21

Applicant: salesforce.com, inc.

Inventor： Ashish Bharadwaj Srinivasa ,  Jacob Alexander Mannix ,  Mario Sergio Rodriguez

IPC: G06F16/2457 ,  G06F16/28

Abstract: Methods, computer readable media, and devices for modeling heterogeneous feature sets for use in personalized search are provided. The method may include generating a similarity factor for each of a plurality of personalization features. For each of the plurality of personalization features, a personalization feature weight is calculated. Each personalization feature weight is converted into a probability distribution and each similarity factor is scaled based on a corresponding probability distribution. Based on each scaled similarity factor, a most recently used affinity value is generated for each corresponding personalization feature. The most recently used affinity values are used to generate a ranking function for use as part of personalized search.