# US 2022/0044137 A1 · Method and apparatus for object preference prediction, and computer readable medium

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

- Number: US 2022/0044137 A1
- Document ID: US20220044137A1
- Status: Published application
- Kind: A1
- Application number: 17/419,152
- Filed: 2019-12-17
- Published: 2022-02-10
- Earliest claimed: 2018-12-29
- Assignees: Beijing Century Tal Education Technology Co., Ltd.
- Applicants: Beijing Century Tal Education Technology Co., Ltd.
- Inventors: Tianqiao LIU; Zitao LIU; Songfan YANG; Yan HUANG; Bangxin ZHANG
- Web page: https://patentgrep.com/patent/US20220044137A1

## CPC

- G06N 3/082: modifying the architecture, e.g. adding, deleting or silencing nodes or connections
- G06N 5/04: Inference or reasoning models
- G06F 16/9535: Search customisation based on user profiles and personalisation
- G06F 17/16: Matrix or vector computation {, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization (matrix transposition G06F7/78)}
- G06N 3/045: Combinations of networks
- G06N 3/0455: Auto-encoder networks; Encoder-decoder networks
- G06N 3/09: Supervised learning
- G06N 5/01: Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound
- G06Q 30/02: Marketing; Price estimation or determination; Fundraising
- G06N 5/02: Knowledge representation; Symbolic representation
- G06Q 30/0631: {Recommending goods or services}
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- 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

The present disclosure relates to a method and an apparatus for object preference prediction, and a computer readable medium. The method includes: acquiring evaluation information indicating preference values of partial users in a user set for partial objects in an object set; acquiring auxiliary information of at least one of the user set and the object set, wherein the auxiliary information indicates an attribute of at least one of a corresponding user in the user set and a corresponding object in the object set; determining a user feature representation and an object feature representation using a matrix decomposition model, based on the evaluation information and the auxiliary information; and determining a preference prediction value of a target user in the user set for a target object in the object set based on the user feature representation and the object feature representation.

## Claims

### Claim 1

A method for object preference prediction, comprising:
acquiring evaluation information indicating preference values of partial users in a user set for partial objects in an object set;
acquiring auxiliary information of at least one of the user set and the object set, wherein the auxiliary information indicates an attribute of at least one of a corresponding user in the user set and a corresponding object in the object set;
determining a user feature representation and an object feature representation using a matrix decomposition model, based on the evaluation information and the auxiliary information; and
determining a preference prediction value of a target user in the user set for a target object in the object set based on the user feature representation and the object feature representation, wherein a preference value of the target user for the target object is not indicated in the evaluation information,
wherein the auxiliary information comprises at least one of object non-hierarchical information, object hierarchical information, user non-hierarchical information and user hierarchical information,
the object non-hierarchical information indicates non-hierarchical attributes of the respective objects in the object set, the object hierarchical information indicates a plurality of attributes, which have a hierarchical relationship, of the respective objects in the object set, the user non-hierarchical information indicates non-hierarchical attributes of the respective users in the user set, and the user hierarchical information indicates a plurality of attributes, which have a hierarchical relationship, of the respective users in the user set.

### Claim 2 (depends on 1)

The method according to claim 1, wherein the auxiliary information comprises the object non-hierarchical information, and wherein the determining the user feature representation and the object feature representation, comprises:
initializing the object feature representation to obtain a current object feature representation;
acquiring a projection of the current object feature representation onto a feature dimension associated with the object non-hierarchical information; and
adjusting the current object feature representation based on a difference between the projection of the current object feature representation and the object non-hierarchical information, to obtain the object feature representation.

### Claim 3 (depends on 1)

The method according to claim 1, wherein the auxiliary information comprises the object hierarchical information, the plurality of attributes of the respective objects are constructed as a plurality of object hierarchies, and wherein the determining the object feature representation, comprises:
determining a plurality of object hierarchical relationship feature representations corresponding to the plurality of object hierarchies based on the object hierarchical information; and
determining the object feature representation based on the plurality of object hierarchical relationship feature representations.

### Claim 4 (depends on 3)

The method according to claim 3, wherein the determining the plurality of object hierarchical relationship feature representations corresponding to the plurality of object hierarchies, comprises:
initializing the plurality of object hierarchical relationship feature representations to obtain a current plurality of object hierarchical relationship feature representations;
determining a first object hierarchy feature representation associated with a first object hierarchy of the plurality of object hierarchies, based on the current plurality of object hierarchical relationship feature representations;
determining a second object hierarchy feature representation associated with a second object hierarchy of the plurality of object hierarchies, based on the current plurality of object hierarchical relationship feature representations, wherein the second object hierarchy is a sub-hierarchy of the first object hierarchy;
determining a prediction object hierarchy feature representation associated with the first object hierarchy, based on the second object hierarchy feature representation and the object hierarchical information; and
adjusting the current plurality of object hierarchical relationship feature representations based on a difference between the first object hierarchy feature representation and the object prediction hierarchy feature representation, to obtain the plurality of object hierarchical relationship feature representations corresponding to the plurality of object hierarchies.

### Claim 5 (depends on 1)

The method according to claim 1, wherein the auxiliary information comprises the user non-hierarchical information, and wherein the determining the user feature representation and the object feature representation, comprises:
initializing the user feature representation to obtain a current user feature representation;
acquiring a projection of the current user feature representation onto a feature dimension associated with the user non-hierarchical information; and
adjusting the current user feature representation based on a difference between the projection of the current user feature representation and the user non-hierarchical information, to obtain the user feature representation.

### Claim 6 (depends on 1)

The method according to claim 1, wherein the auxiliary information comprises the user hierarchical information, the plurality of attributes of the respective users are constructed as a plurality of user hierarchies, and wherein the determining the user feature representation and the object feature representation, comprises:
determining a plurality of user hierarchical relationship feature representations corresponding to the plurality of user hierarchies based on the user hierarchical information; and
determining the user feature representation based on the plurality of user hierarchical relationship feature representations.

### Claim 7 (depends on 6)

The method according to claim 6, wherein the determining the plurality of user hierarchical relationship feature representations corresponding to the plurality of user hierarchies, comprises:
initializing the plurality of user hierarchical relationship feature representations to obtain a current plurality of user hierarchical relationship feature representations;
determining a first user hierarchy feature representation associated with a first user hierarchy of the plurality of user hierarchies, based on the current plurality of user hierarchical relationship feature representations;
determining a second user hierarchy feature representation associated with a second user hierarchy of the plurality of user hierarchies, based on the current plurality of user hierarchical relationship feature representations, wherein the second user hierarchy is a sub-hierarchy of the first user hierarchy;
determining a prediction user hierarchy feature representation associated with the first user hierarchy, based on the second user hierarchy feature representation and the user hierarchical information; and
adjusting the current plurality of user hierarchical relationship feature representations based on a difference between the first user hierarchy feature representation and the user prediction hierarchy feature representation, to obtain the plurality of user hierarchical relationship feature representations corresponding to the plurality of user hierarchies.

### Claim 8 (depends on 1)

The method according to claim 1, wherein the determining the preference prediction value of the target user in the user set for the target object in the object set, comprises:
calculating a matrix product of the user feature representation and the object feature representation; and
determining the preference prediction value of the target user for the target object based on the matrix product.

### Claim 9 (depends on 1)

The method according to claim 1, further comprising:
recommending an object in the object set to the target user, based on the evaluation information and the preference prediction value.

### Claim 10

An apparatus for object preference prediction, comprising:
an evaluation information acquiring module, configured to acquire evaluation information indicating preference values of partial users in a user set for partial objects in an object set;
an auxiliary information acquiring module, configured to acquire auxiliary information of at least one of the user set and the object set, wherein the auxiliary information indicates an attribute of at least one of a corresponding user in the user set and a corresponding object in the object set;
a feature representation determining module, configured to determine a user feature representation and an object feature representation using a matrix decomposition model, based on the evaluation information and the auxiliary information; and
a preference prediction determining module, configured to determine a preference prediction value of a target user in the user set for a target object in the object set based on the user feature representation and the object feature representation, wherein a preference value of the target user for the target object is not indicated in the evaluation information,
wherein the auxiliary information comprises at least one of object non-hierarchical information, object hierarchical information, user non-hierarchical information and user hierarchical information,
the object non-hierarchical information indicates non-hierarchical attributes of the respective objects in the object set, the object hierarchical information indicates a plurality of attributes, which have a hierarchical relationship, of the respective objects in the object set, the user non-hierarchical information indicates non-hierarchical attributes of the respective users in the user set, and the user hierarchical information indicates a plurality of attributes, which have a hierarchical relationship, of the respective users in the user set.

### Claim 11 (depends on 10)

The apparatus according to claim 10, wherein the auxiliary information comprises the object non-hierarchical information, and wherein the feature representation determining module comprises:
a first initialization module, configured to initialize the object feature representation to obtain a current object feature representation;
a first projection acquiring module, configured to acquire a projection of the current object feature representation onto a feature dimension associated with the object non-hierarchical information; and
a first adjustment module, configured to adjust the current object feature representation based on a difference between the projection of the current object feature representation and the object non-hierarchical information, to obtain the object feature representation.

### Claim 12 (depends on 10)

The apparatus according to claim 10, wherein the auxiliary information comprises the object hierarchical information, the plurality of attributes of the respective objects are constructed as a plurality of object hierarchies, and wherein the feature representation determining module comprises:
an object hierarchical relationship feature representation determining module, configured to determine a plurality of object hierarchical relationship feature representations corresponding to the plurality of object hierarchies based on the object hierarchical information; and
an object feature representation determining module, configured to determine the object feature representation based on the plurality of object hierarchical relationship feature representations.

### Claim 13 (depends on 12)

The apparatus according to claim 12, wherein the object hierarchical relationship feature representation determining module comprises:
a second initialization module, configured to initialize the plurality of object hierarchical relationship feature representations to obtain a current plurality of object hierarchical relationship feature representations;
a first object hierarchy feature representation determining module, configured to determine a first object hierarchy feature representation associated with a first object hierarchy of the plurality of object hierarchies, based on the current plurality of object hierarchical relationship feature representations;
a second object hierarchy feature representation determining module, configured to determine a second object hierarchy feature representation associated with a second object hierarchy of the plurality of object hierarchies, based on the current plurality of object hierarchical relationship feature representations, wherein the second object hierarchy is a sub-hierarchy of the first object hierarchy;
a prediction object hierarchy representation determining module, configured to determine a prediction object hierarchy feature representation associated with the first object hierarchy, based on the second object hierarchy feature representation and the object hierarchical information; and
a second adjustment module, configured to adjust the current plurality of object hierarchical relationship feature representations based on a difference between the first object hierarchy feature representation and the object prediction hierarchy feature representation, to obtain the plurality of object hierarchical relationship feature representations corresponding to the plurality of object hierarchies.

### Claim 14 (depends on 10)

The apparatus according to claim 10, wherein the auxiliary information comprises the user non-hierarchical information, and wherein the feature representation determining the module comprises:
a third initialization module, configured to initialize the user feature representation to obtain a current user feature representation;
a second projection acquiring module, configured to acquire a projection of the current user feature representation onto a feature dimension associated with the user non-hierarchical information; and
a third adjustment module, configured to adjust the current user feature representation based on a difference between the projection of the current user feature representation and the user non-hierarchical information, to obtain the user feature representation.

### Claim 15 (depends on 10)

The apparatus according to claim 10, wherein the auxiliary information comprises the user hierarchical information, the plurality of attributes of the respective users are constructed as a plurality of user hierarchies, and wherein the feature representation determining module comprises:
a user hierarchical relationship feature representation determining module, configured to determine a plurality of user hierarchical relationship feature representations corresponding to the plurality of user hierarchies based on the user hierarchical information; and
a user feature representation determining module, configured to determine the user feature representation based on the plurality of user hierarchical relationship feature representations.

### Claim 16 (depends on 15)

The apparatus according to claim 15, wherein the user hierarchical relationship feature representation determining module comprises:
a fourth initialization module, configured to initialize the plurality of user hierarchical relationship feature representations to obtain a current plurality of user hierarchical relationship feature representations;
a first user hierarchy feature representation determining module, configured to determine a first user hierarchy feature representation associated with a first user hierarchy of the plurality of user hierarchies, based on the current plurality of user hierarchical relationship feature representations;
a second user hierarchy feature representation determining module, configured to determine a second user hierarchy feature representation associated with a second user hierarchy of the plurality of user hierarchies, based on the current plurality of user hierarchical relationship feature representations, wherein the second user hierarchy is a sub-hierarchy of the first user hierarchy;
a prediction user hierarchy representation determining module, configured to determine a prediction user hierarchy feature representation associated with the first user hierarchy, based on the second user hierarchy feature representation and the user hierarchical information; and
a fourth adjustment module, configured to adjust the current plurality of user hierarchical relationship feature representations based on a difference between the first user hierarchy feature representation and the user prediction hierarchy feature representation, to obtain the plurality of user hierarchical relationship feature representations corresponding to the plurality of user hierarchies.

### Claim 17 (depends on 10)

The apparatus according to claim 10, wherein the preference prediction determining module comprises:
a matrix product calculation module, configured to calculate a matrix product of the user feature representation and the object feature representation; and
a preference prediction value determining module, configured to determine the preference prediction value of the target user for the target object based on the matrix product.

### Claim 18 (depends on 10)

The apparatus according to claim 10, further comprising:
a recommendation module, configured to recommend an object in the object set to the target user, based on the evaluation information and the preference prediction value.

### Claim 19 (depends on 1)

A computer readable storage medium having computer readable program instructions stored thereon for executing the method according to claim 1.