# US 2019/0377768 A1 · Application recommendation method and apparatus, and server

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

- Number: US 2019/0377768 A1
- Document ID: US20190377768A1
- Status: Published application
- Kind: A1
- Application number: 16/550,405
- Filed: 2019-08-26
- Published: 2019-12-12
- Earliest claimed: 2017-05-11
- Assignees: Tencent Technology (Shenzhen) Company Limited
- Applicants: Tencent Technology (Shenzhen) Company Limited
- Inventors: Xing Ya WANG; Wei WAN
- Web page: https://patentgrep.com/patent/US20190377768A1

## CPC

- G06F 16/9536: Search customisation based on social or collaborative filtering
- G06N 5/022: Knowledge engineering; Knowledge acquisition
- G06N 7/01: Probabilistic graphical models, e.g. probabilistic networks
- G06N 20/00: Machine learning
- G06Q 50/00: Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism (healthcare informatics G16H)
- H04L 51/52: for supporting social networking services
- 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
- H04L: TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION (arrangements common to telegraphic and telephonic communication H04M)

## Abstract

A method, apparatus and server for recommending an application are provided. The method may include determining at least one target friend among a plurality of friends of a target user, obtaining social interaction characteristics between the target user and the at least one target friend, based on a pre-trained friend impact prediction model of a target application and the social interaction characteristics between the target user and the at least one target friend, determining a first impact of the at least one target friend on registration of the target user in the target application, determining a recommendation score corresponding to the target user and the target application according to the first impact of the at least one target friend on registration of the target user in the target application, and performing application recommendation according to the recommendation score.

## Claims

### Claim 1

An application recommendation method, the method being performed by a server, and the method comprising:
determining at least one target friend among a plurality of friends of a target user;
obtaining social interaction characteristics between the target user and the at least one target friend;
based on a pre-trained friend impact prediction model of a target application and the social interaction characteristics between the target user and the at least one target friend, determining a first impact of the at least one target friend on registration of the target user in the target application;
determining a recommendation score corresponding to the target user and the target application according to the first impact of the at least one target friend on registration of the target user in the target application; and
performing application recommendation according to the recommendation score.

### Claim 2 (depends on 1)

The application recommendation method according to claim 1, further comprising:
sampling a preset quantity of a plurality of registration users from users having registered in the target application;
selecting, among the preset quantity of registration users, a friend registered with the target application as a positive sample, and selecting a friend not registered with the target application as a negative sample;
obtaining, for each one of the plurality of registration users, positive sample characteristics based on social interaction characteristics between one of the plurality of registration users and at least one positive sample;
obtaining, for each one of the plurality of registration users, negative sample characteristics based on social interaction characteristics between the one of the plurality of registration users and at least one negative sample; and
training the positive sample characteristics and the negative sample characteristics corresponding to the plurality of registration users according to a machine learning algorithm, to obtain the pre-trained friend impact prediction model of the target application.

### Claim 3 (depends on 1)

The application recommendation method according to claim 1, wherein the social interaction characteristic between the target user and the at least one target friend comprises:
obtaining a set of social interaction behavior types; and
determining a quantity of social interaction behaviors based on the set of social interaction behavior types within a set interaction period.

### Claim 4 (depends on 3)

The application recommendation method according to claim 3, wherein the pre-trained friend impact prediction model determines weights for each one of the set social interaction behavior types, and
wherein the determining the first impact further comprises:
combining, for the at least one target friend of the target user, the quantity of social interaction behaviors belonging to the set of social interaction behavior types with the weights corresponding to the social interaction behavior types; and
determining the first impact of the at least one target friend on the registration of the target user in the target application.

### Claim 5 (depends on 1)

The application recommendation method according to claim 1, wherein the determining the recommendation score further comprises:
determining a first sum of impacts of the target friends by summing a plurality of first impacts of the target friends on the registration of the target user in the target application; and
using the first sum of impacts of the target friends as the recommendation score corresponding to the target user and the target application.

### Claim 6 (depends on 1)

The application recommendation method according to claim 1, further comprising:
determining at least one friend circle of the target user;
determining a second impact of the at least one friend circle on the registration of the target user in the target application; and
determining the recommendation score corresponding to the target user and the target application according to the first impact of the target friends on the registration of the target user in the target application and the second impact of the friend circles on the registration of the target user in the target application.

### Claim 7 (depends on 6)

The application recommendation method according to claim 6, wherein the method further comprises:
determining a first sum of impacts of the target friends according to the first impacts of the target friends on the registration of the target user in the target application and a second sum of impacts according to the second impact of the at least one friend circle on the registration of the target user in the target application; and
adding the first sum of the impacts and the second sum of the impacts to obtain the recommendation score corresponding to the target user and the target application.

### Claim 8 (depends on 5)

The application recommendation method according to claim 5, wherein the determining the first sum of impacts of the target friends further comprises:
based on the at least one target friend being a friend registered in the target application, adding the plurality of first impacts of the target friends on registration of the target user in the target application, and obtaining the first sum of the impacts of the target friends; or
based on the at least one target friend being a friend registered in the target application and a friend not registered in the target application, combining the plurality of first impacts of the target friends on registration of the target user in the target application corresponding to scoring thresholds, and adding the combined result to obtain the first sum of the impacts of the target friends,
wherein the scoring thresholds comprise a first value and a second value, the first value is set for the target friend registered in the target application, and the second value is set for the target friend not registered in the target application.

### Claim 9 (depends on 7)

The application recommendation method according to claim 7, wherein the determining the second sum of impacts of the friend circles further comprises:
adding a plurality of the second impacts of the friend circles on registration of the target user in the target application, and obtaining the second sum of the impacts of the friend circles on registration of the target user in the target application.

### Claim 10 (depends on 6)

The application recommendation method according to claim 6, wherein the determining at least one friend circle of the target user comprises:
determining direct friends of the target user;
rejecting a direct friend having no mutual friend, and obtaining a friend range of a to-be-divided friend circle corresponding to the target user;
setting friends in the friend range as nodes, and connecting the nodes according to social relationships between the direct friends in the friend range;
designating relationship labels for the nodes corresponding to the direct friends in the friend range according to the social interaction characteristics between the target user and the direct friends in the friend range; and
refreshing relationship labels of all nodes each round until a predetermined refreshing convergence condition is satisfied, to obtain a node cluster, wherein each round of refreshing process determines, for any one node, relationship labels of neighboring nodes of the node, and designating the node with a relationship label having a largest quantity.

### Claim 11 (depends on 6)

The application recommendation method according to claim 6, wherein the determining the second impact of the at least one friend circle on registration of the target user in the target application further comprises:
determining a friend registered in the target application in the at least one target friend of the target user, and determining the friend circle of the target user in which the at least one target friend having registered in the target application is located; and
determining, for any one of a plurality of friend circles of the target user, the second impact of the friend circle on the registration of the target user in the target application according to a quantity of friends in the friend circle and a quantity of friends in the friend circle having registered in the target application,
wherein the one of the plurality of friend circles is distinguished from the other one of the plurality of friend circles based on a friend circle class.

### Claim 12 (depends on 11)

The application recommendation method according to claim 11, further comprising:
determining, for at least one first friend circle in which group chats established, a quantity of communication information based on a communication threshold;
determining, in the group chats established in the first friend circle, a group chat in which members of the first friend circle occupies a proportion greater than a predetermined proportion;
collecting a name text of the determined group chat, and performing text classification on the name text, and obtaining a class of the first friend circle; and
determining, according to pre-trained prediction models of friend circle classes and a circle characteristic of a friend circle of a to-be-determined friend circle class, prediction results of the prediction models of the friend circle classes, and determining a friend circle class having a maximum probability in the prediction results as the friend circle class of the friend circle, wherein the prediction model of one friend circle class is obtained through training according to the circle characteristic of the friend circle class of the friend circle and a circle characteristic of the friend circle being not of the friend circle class.

### Claim 13 (depends on 1)

The method according to claim 1, wherein the performing application recommendation according to the recommendation score comprises:
ranking a plurality of target applications according to the recommendation score corresponding to the target user, and recommending, according to an application ranking result, a top-ranked target application to the target user; or
ranking a plurality of target users according to the recommendation score corresponding to the target application, and recommending the target application to a top-ranked target user according to a user ranking result.

### Claim 14

An application recommendation apparatus, comprising:
at least one memory configured to store computer program code; and
at least one processor configured to access the computer program code comprising:
friend determining code configured to cause the at least one processor to determine at least one target friend among a plurality of friends of a target user;
interaction characteristic obtaining code configured to cause the at least one processor to obtain social interaction characteristics between the target user and the at least one target friend;
first impact determining code configured to cause the at least one processor to determine, based on a pre-trained friend impact prediction model of a target application and social interaction characteristics between the target user and the at least one target friend, a first impact of the at least one target friend on registration of the target user in the target application;
recommendation score determining code configured to cause the at least one processor to determine a recommendation score corresponding to the target user and the target application according to the first impact of the at least one target friend on registration of the target user in the target application; and
recommendation code configured to cause the at least one processor to perform application recommendation according to the recommendation score.

### Claim 15

An application recommendation server, comprising:
at least one memory configured to store computer program code; and
at least one processor configured to access the at least one memory and operate according to the computer program code comprising:
friend determining code configured to cause the at least one processor to determine at least one target friend among a plurality of friends of a target user;
interaction characteristic obtaining code configured to cause the at least one processor to obtain social interaction characteristics between the target user and the at least one target friend;
first impact determining code configured to cause the at least one processor to determine, based on a pre-trained friend impact prediction model of a target application and the social interaction characteristics between the target user and the at least one target friend, a first impact of the at least one target friend on registration of the target user in the target application;
recommendation score determining code configured to cause the at least one processor to determine a recommendation score corresponding to the target user and the target application according to the first impact of the at least one target friend on registration of the target user in the target application; and
recommendation code configured to cause the at least one processor to perform application recommendation according to the recommendation score.

### Claim 16 (depends on 1)

A non-transitory computer readable storage medium, storing executable instructions, the executable instructions capable of causing a computer to perform the application recommendation method according to claim 1.

### Claim 17 (depends on 14)

The application recommendation apparatus according to claim 14, wherein the interaction characteristic obtaining code is further configured to cause the at least one processor to:
obtain a set of social interaction behavior types; and
determine a quantity of social interaction behaviors based on the set of social interaction behavior types within a set interaction period.

### Claim 18 (depends on 14)

The application recommendation apparatus according to claim 14, further comprising:
friend circle determining code configured to cause the at least one processor to determine at least one friend circle of the target user;
second impact determining code configured to cause the at least one processor to determine a second impact of the at least one friend circle on registration of the target user in the target application; and
recommendation score determining code further configured to cause the at least one processor to determine the recommendation score corresponding to the target user and the target application according to the first impact of the target friends on the registration of the target user in the target application and the second impact of the friends circle on the registration of the target user in the target application.

### Claim 19 (depends on 15)

The application recommendation server according to claim 15, wherein the interaction characteristic obtaining code is further configured to cause the at least one processor to:
obtain a set of social interaction behavior types; and
determine a quantity of social interaction behaviors based on the set of social interaction behavior types within a set interaction period.

### Claim 20 (depends on 15)

The application recommendation server according to claim 15, further comprising:
friend circle determining code configured to cause the at least one processor to determine at least one friend circle of the target user;
second impact determining code configured to cause the at least one processor to determine a second impact of the at least one friend circle on registration of the target user in the target application; and
recommendation score determining code further configured to cause the at least one processor to determine the recommendation score corresponding to the target user and the target application according to the first impact of the target friends on the registration of the target user in the target application and the second impact of the friends circle on the registration of the target user in the target application.
