# US 2021/0173884 A1 · Recommendation method and apparatus, and storage medium

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2021/0173884 A1
- Document ID: US20210173884A1
- Status: Published application
- Kind: A1
- Application number: 17/183,251
- Filed: 2021-02-23
- Published: 2021-06-10
- Earliest claimed: 2018-11-29
- Assignees: Tencent Technology (Shenzhen) Company Limited
- Applicants: Tencent Technology (Shenzhen) Company Limited
- Inventors: Xu ZHANG; Leyu LIN; Kaikai GE; Linyao TANG; Yudan LIU; Xin CHEN; Su YAN; Kai ZHUANG; Wei WANG; Jing ZHANG
- Web page: https://patentgrep.com/patent/US20210173884A1

## CPC

- G06F 16/9536: Search customisation based on social or collaborative filtering
- G06F 16/9535: Search customisation based on user profiles and personalisation
- G06N 3/045: Combinations of networks
- G06N 3/08: Learning methods
- G06N 3/09: Supervised learning
- G06Q 10/42: Determination of affinities or common interests between users
- G06Q 10/46: Determination of level of influence of users within social networking services
- G06N 3/048: Activation functions
- G06N 5/022: Knowledge engineering; Knowledge acquisition
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- 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

## Abstract

A recommendation method is provided. In the method, a candidate item to be recommended to a social network user is obtained. The social network user has at least two different types of social relationships. For at least one target social object in each of the at least two different types of social relationships of the social network user, attention of each of the at least one target social object in the respective type of social relationship to the candidate item is determined. According to the attention of each of the at least one target social object in the at least two different types of social relationships to the candidate item, a comprehensive attention of the target social objects of the at least two different types of social relationships to the candidate item is determined. According to the comprehensive attention, whether to recommend the candidate item to the social network user is determined.

## Claims

### Claim 1

A recommendation method, comprising:
obtaining a candidate item to be recommended to a social network user, the social network user having at least two different types of social relationships;
determining, for at least one target social object in each of the at least two different types of social relationships of the social network user, attention of each of the at least one target social object in the respective type of social relationship to the candidate item;
determining, by processing circuitry and according to the attention of each of the at least one target social object in the at least two different types of social relationships to the candidate item, a comprehensive attention of the target social objects of the at least two different types of social relationships to the candidate item; and
determining, according to the comprehensive attention, whether to recommend the candidate item to the social network user.

### Claim 2 (depends on 1)

The method according to claim 1, wherein the at least two different types of social relationships includes a group relationship, and the at least one target social object of the group relationship type includes at least one group joined by the social network user on a social network.

### Claim 3 (depends on 2)

The method according to claim 2, wherein the determining the attention comprises:
inputting a vector representation of the at least one group and a vector representation of the candidate item to a first attention model, the first attention model being pre-trained with an attention parameter of each group on the social network to the candidate item; and
obtaining attention information of each group to the candidate item that is outputted by the first attention model, the attention information of each group indicating an attention of each group to the candidate item that is generated and outputted by the first attention model according to an attention parameter of the respective group to the candidate item determined according to the vector representation of the respective group and the vector representation of the candidate item.

### Claim 4 (depends on 3)

The method according to claim 3, wherein for each group, the first attention model is pre-trained with an importance weight of each group member in the each group; and
the attention information of each group to the candidate item that is outputted by the first attention model is obtained by further performing weighting processing on the importance weight of the each group member in the respective group and the attention parameter of the respective group to the candidate item.

### Claim 5 (depends on 2)

The method according to claim 2, wherein the at least two different types of social relationships includes a friend relationship, and the at least one target social object of the friend relationship type includes at least one friend added by the social network user on the social network.

### Claim 6 (depends on 5)

The method according to claim 5, wherein the determining the attention further comprises:
inputting a vector representation of the at least one friend and the vector representation of the candidate item to a second attention model, the second attention model being pre-trained with an attention parameter of each friend on the social network to the candidate item; and
obtaining attention information of each friend to the candidate item that is outputted by the second attention model, the attention information of each friend indicating an attention of each friend to the candidate item that is generated by the second attention model according to an attention parameter of the respective friend to the candidate item determined according to the vector representation of the respective friend and the vector representation of the candidate item.

### Claim 7 (depends on 1)

The method according to claim 1, wherein
the determining the comprehensive attention includes:
performing normalization processing on the attention of each target social object to the candidate item; and
performing summation processing on attention of target social objects of the same type to the candidate item after the normalization processing, to obtain a comprehensive attention of the target social objects of each type to the candidate item; and
the determining whether to recommend includes:
determining a recommendation index for recommending the candidate item based on the comprehensive attention of social objects of each type to the candidate item; and
determining, according to the recommendation index, whether to recommend the candidate item to the social network user.

### Claim 8

A recommendation apparatus, comprising:
processing circuitry configured to:
obtain a candidate item to be recommended to a social network user, the social network user having at least two different types of social relationships;
determine, for at least one target social object in each of the at least two different types of social relationships of the social network user, attention of each of the at least one target social object in the respective type of social relationship to the candidate item;
determine, according to the attention of each of the at least one target social object in the at least two different types of social relationships to the candidate item, a comprehensive attention of the target social objects of the at least two different types of social relationships to the candidate item; and
determine, according to the comprehensive attention, whether to recommend the candidate item to the social network user.

### Claim 9 (depends on 8)

The recommendation apparatus according to claim 8, wherein the at least two different types of social relationships includes a group relationship, and the at least one target social object of the group relationship type includes at least one group joined by the social network user on a social network.

### Claim 10 (depends on 9)

The recommendation apparatus according to claim 9, wherein the processing circuitry is configured to:
input a vector representation of the at least one group and a vector representation of the candidate item to a first attention model, the first attention model being pre-trained with an attention parameter of each group on the social network to the candidate item; and
obtain attention information of each group to the candidate item that is outputted by the first attention model, the attention information of each group indicating an attention of each group to the candidate item that is generated and outputted by the first attention model according to an attention parameter of the respective group to the candidate item determined according to the vector representation of the respective group and the vector representation of the candidate item.

### Claim 11 (depends on 10)

The recommendation apparatus according to claim 10, wherein for each group, the first attention model is pre-trained with an importance weight of each group member in the each group; and
the attention information of each group to the candidate item that is outputted by the first attention model is obtained by further performing weighting processing on the importance weight of the each group member in the respective group and the attention parameter of the respective group to the candidate item.

### Claim 12 (depends on 9)

The recommendation apparatus according to claim 9, wherein the at least two different types of social relationships includes a friend relationship, and the at least one target social object of the friend relationship type includes at least one friend added by the social network user on the social network.

### Claim 13 (depends on 12)

The recommendation apparatus according to claim 12, wherein the processing circuitry is configured to:
input a vector representation of the at least one friend and the vector representation of the candidate item to a second attention model, the second attention model being pre-trained with an attention parameter of each friend on the social network to the candidate item; and
obtain attention information of each friend to the candidate item that is outputted by the second attention model, the attention information of each friend indicating an attention of each friend to the candidate item that is generated by the second attention model according to an attention parameter of the respective friend to the candidate item determined according to the vector representation of the respective friend and the vector representation of the candidate item.

### Claim 14 (depends on 8)

The recommendation apparatus according to claim 8, wherein the processing circuitry is configured to:
perform normalization processing on the attention of each target social object to the candidate item;
perform summation processing on attention of target social objects of the same type to the candidate item after the normalization processing, to obtain a comprehensive attention of the target social objects of each type to the candidate item;
determine a recommendation index for recommending the candidate item based on the comprehensive attention of social objects of each type to the candidate item; and
determine, according to the recommendation index, whether to recommend the candidate item to the social network user.

### Claim 15

A non-transitory computer-readable storage medium storing instructions which when executed by at least one processor cause the at least one processor to perform:
obtaining a candidate item to be recommended to a social network user, the social network user having at least two different types of social relationships;
determining, for at least one target social object in each of the at least two different types of social relationships of the social network user, attention of each of the at least one target social object in the respective type of social relationship to the candidate item;
determining, according to the attention of each of the at least one target social object in the at least two different types of social relationships to the candidate item, a comprehensive attention of the target social objects of the at least two different types of social relationships to the candidate item; and
determining, according to the comprehensive attention, whether to recommend the candidate item to the social network user.

### Claim 16 (depends on 15)

The non-transitory computer-readable storage medium according to claim 15, wherein the at least two different types of social relationships includes a group relationship, and the at least one target social object of the group relationship type includes at least one group joined by the social network user on a social network.

### Claim 17 (depends on 16)

The non-transitory computer-readable storage medium according to claim 16, wherein the determining the attention comprises:
inputting a vector representation of the at least one group and a vector representation of the candidate item to a first attention model, the first attention model being pre-trained with an attention parameter of each group on the social network to the candidate item; and
obtaining attention information of each group to the candidate item that is outputted by the first attention model, the attention information of each group indicating an attention of each group to the candidate item that is generated and outputted by the first attention model according to an attention parameter of the respective group to the candidate item determined according to the vector representation of the respective group and the vector representation of the candidate item.

### Claim 18 (depends on 17)

The non-transitory computer-readable storage medium according to claim 17, wherein for each group, the first attention model is pre-trained with an importance weight of each group member in the each group; and
the attention information of each group to the candidate item that is outputted by the first attention model is obtained by further performing weighting processing on the importance weight of the each group member in the respective group and the attention parameter of the respective group to the candidate item.

### Claim 19 (depends on 16)

The non-transitory computer-readable storage medium according to claim 16, wherein the at least two different types of social relationships includes a friend relationship, and the at least one target social object of the friend relationship type includes at least one friend added by the social network user on the social network.

### Claim 20 (depends on 19)

The non-transitory computer-readable storage medium according to claim 19, wherein the determining the attention further comprises:
inputting a vector representation of the at least one friend and the vector representation of the candidate item to a second attention model, the second attention model being pre-trained with an attention parameter of each friend on the social network to the candidate item; and
obtaining attention information of each friend to the candidate item that is outputted by the second attention model, the attention information of each friend indicating an attention of each friend to the candidate item that is generated by the second attention model according to an attention parameter of the respective friend to the candidate item determined according to the vector representation of the respective friend and the vector representation of the candidate item.