CN-122019865-A - Information processing method, apparatus, electronic device, storage medium, and program product

CN122019865ACN 122019865 ACN122019865 ACN 122019865ACN-122019865-A

Abstract

The embodiment of the application discloses an information processing method, an information processing device, electronic equipment, a storage medium and a program product, wherein the information processing method comprises the steps of obtaining a plurality of pieces of characteristic information of attribute information of an object to be processed, determining characteristic information quantity of the characteristic information, carrying out grouping processing on the plurality of pieces of characteristic information according to the characteristic information quantity to obtain a plurality of characteristic groups, carrying out multiple cumulative merging on the plurality of characteristic groups to obtain a plurality of hierarchical structures arranged according to a hierarchy, and carrying out embedded representation learning on the basis of the hierarchical structures in adjacent two hierarchical structures of the plurality of hierarchical structures, wherein the next hierarchical structure comprises the characteristic groups in the last hierarchical structure, so as to obtain embedded representation of the attribute information of the object to be processed, and carrying out content recommendation on the object to be processed on the basis of the embedded representation. The scheme can effectively promote the learning effect of embedded representation learning.

Inventors

YAN JUNCHAO
CHEN YIMING
CHEN RONG
FAN GUOTAO
ZHANG SHUBIN
WANG LIFENG

Assignees

腾讯科技（深圳）有限公司

Dates

Publication Date: 20260512
Application Date: 20241112

Claims (11)

1. An information processing method, characterized by comprising: acquiring a plurality of characteristic information of attribute information of an object to be processed; Determining the characteristic information quantity of the characteristic information, and carrying out grouping processing on the plurality of characteristic information according to the characteristic information quantity to obtain a plurality of characteristic groups; Accumulating and merging the feature groups for multiple times to obtain a plurality of hierarchical structures arranged according to a hierarchy, wherein the next hierarchical structure comprises the feature groups in the previous hierarchical structure in two adjacent hierarchical structures of the plurality of hierarchical structures; And performing embedded representation learning based on the hierarchical structure to obtain embedded representation of the attribute information of the object to be processed, so as to recommend content to the object to be processed based on the embedded representation.
2. The method of claim 1, wherein the performing a plurality of cumulative combinations of the plurality of feature sets to obtain a plurality of hierarchical structures arranged in a hierarchy comprises: determining a group information amount of each of the feature groups; sorting the feature groups of the feature groups according to the order of the group information quantity from large to small to obtain a plurality of sorted feature groups; starting from the first feature group in the sorted feature groups, carrying out accumulation and combination on the sorted feature groups for a plurality of times to obtain the hierarchical structures.
3. The method of claim 2, wherein starting from a first feature set of the ordered plurality of feature sets, performing a plurality of cumulative merges on the ordered plurality of feature sets to obtain the plurality of hierarchies, comprising: determining a first feature group in the sorted feature groups as a first hierarchical structure; combining the first feature group with a next feature group of the first feature group in the sequenced plurality of feature groups to obtain a next hierarchical structure of the first hierarchical structure; And taking the next hierarchical structure as a new first feature group, and returning to execute the step of merging the first feature group with the next feature group of the first feature group in the sorted multiple feature groups to obtain the next hierarchical structure of the first hierarchical structure until merging the next feature group into the last feature group in the sorted multiple feature groups to obtain the multiple hierarchical structures.
4. The method according to claim 1, wherein the learning of the embedded representation based on the hierarchical structure to obtain the embedded representation of the attribute information of the object to be processed comprises: respectively carrying out embedded representation learning on each hierarchical structure to obtain a sub-embedded representation corresponding to each hierarchical structure; and generating an embedded representation of the attribute information of the object to be processed based on the sub-embedded representation.
5. The method of claim 4, wherein the learning the embedded representation of each hierarchical structure to obtain the sub-embedded representation corresponding to each hierarchical structure includes: filtering noise characteristic information in the hierarchical structure aiming at each hierarchical structure to obtain a filtered hierarchical structure; And performing embedded representation learning based on the filtered hierarchical structure to obtain a sub-embedded representation corresponding to the hierarchical structure.
6. The method of claim 5, wherein filtering the noise signature information in the hierarchy to obtain a filtered hierarchy comprises: Acquiring a frequency threshold corresponding to the hierarchical structure, wherein the frequency threshold is related to the hierarchy of the hierarchical structure; And filtering the characteristic information with the occurrence frequency smaller than or equal to the frequency threshold value in the hierarchical structure to obtain a filtered hierarchical structure.
7. The method according to any one of claims 1 to 6, wherein the grouping the plurality of feature information according to the feature information amount to obtain a plurality of feature groups includes: Acquiring the grouping number and the total feature number of the plurality of feature information; Calculating and determining the group feature quantity allocated to the feature group according to the total feature quantity and the grouping group quantity; Sequencing the plurality of feature information according to the sequence from the large feature information quantity to the small feature information quantity to obtain sequenced plurality of feature information; and carrying out grouping processing on the sequenced plurality of feature information according to the group feature quantity to obtain a plurality of feature groups, wherein the feature groups comprise feature information corresponding to the group feature quantity.
8. An information processing apparatus, characterized by comprising: an acquisition unit configured to acquire a plurality of pieces of feature information of attribute information of an object to be processed; The grouping unit is used for determining the characteristic information quantity of the characteristic information, and grouping the plurality of characteristic information according to the characteristic information quantity to obtain a plurality of characteristic groups; The merging unit is used for carrying out accumulation and merging on the plurality of feature groups for a plurality of times to obtain a plurality of hierarchical structures which are arranged according to a hierarchy, wherein the next hierarchical structure comprises the feature groups in the previous hierarchical structure in two adjacent hierarchical structures of the plurality of hierarchical structures; and the learning unit is used for carrying out embedded representation learning based on the hierarchical structure to obtain embedded representation of the attribute information of the object to be processed so as to carry out content recommendation on the object to be processed based on the embedded representation.
9. An electronic device, comprising a processor and a memory, wherein the memory stores a plurality of instructions, and wherein the processor loads instructions from the memory to perform the steps in the information processing method according to any one of claims 1 to 7.
10. A computer readable storage medium storing a plurality of instructions adapted to be loaded by a processor for performing the steps of the information processing method according to any one of claims 1 to 7.
11. A computer program product comprising computer programs/instructions which, when executed by a processor, implement the steps of the information processing method of any of claims 1 to 7.

Description

Information processing method, apparatus, electronic device, storage medium, and program product Technical Field The present application relates to the field of machine learning, and in particular, to an information processing method, apparatus, electronic device, storage medium, and program product. Background Embedding (Embedding) is a very important concept in machine learning, which is mainly used to transform high-dimensional discrete variables into low-dimensional continuous vector representations, which is critical to improve model performance, as it can help the model better understand and utilize structured information in the data, and thus is widely used in recommendation systems. Where embedding means learning is a key step to achieve this goal of embedding. However, when learning the embedded representation of the features in the related art, a fixed embedded dimension is often used, so for those features with less information and lower complexity, the high-dimensional embedding does not bring about significant improvement in performance, but increases the storage pressure and the calculation cost during the embedded learning, so that the learning effect of the embedded representation is poor. Disclosure of Invention The embodiment of the application provides an information processing method, an information processing device, electronic equipment, a storage medium and a program product, which can effectively improve the learning effect of embedded representation learning. The embodiment of the application provides an information processing method, which comprises the following steps: acquiring a plurality of characteristic information of attribute information of an object to be processed; Determining the characteristic information quantity of the characteristic information, and carrying out grouping processing on the plurality of characteristic information according to the characteristic information quantity to obtain a plurality of characteristic groups; Accumulating and merging the feature groups for multiple times to obtain a plurality of hierarchical structures arranged according to a hierarchy, wherein the next hierarchical structure comprises the feature groups in the previous hierarchical structure in two adjacent hierarchical structures of the plurality of hierarchical structures; And performing embedded representation learning based on the hierarchical structure to obtain embedded representation of the attribute information of the object to be processed, so as to recommend content to the object to be processed based on the embedded representation. The embodiment of the application also provides an information processing device, which comprises: an acquisition unit configured to acquire a plurality of pieces of feature information of attribute information of an object to be processed; The grouping unit is used for determining the characteristic information quantity of the characteristic information, and grouping the plurality of characteristic information according to the characteristic information quantity to obtain a plurality of characteristic groups; The merging unit is used for carrying out accumulation and merging on the plurality of feature groups for a plurality of times to obtain a plurality of hierarchical structures which are arranged according to a hierarchy, wherein the next hierarchical structure comprises the feature groups in the previous hierarchical structure in two adjacent hierarchical structures of the plurality of hierarchical structures; and the learning unit is used for carrying out embedded representation learning based on the hierarchical structure to obtain embedded representation of the attribute information of the object to be processed so as to carry out content recommendation on the object to be processed based on the embedded representation. In some embodiments, the merging unit comprises: A determining subunit configured to determine a group information amount of each of the feature groups; the sorting subunit is used for sorting the feature groups of the feature groups according to the sequence from the large information quantity of the groups to the small information quantity of the groups to obtain a plurality of sorted feature groups; And the merging subunit is used for starting from the first feature group in the sorted multiple feature groups, and carrying out multiple accumulated merging on the sorted multiple feature groups to obtain the multiple hierarchical structures. In some embodiments, the merging subunit is further configured to: determining a first feature group in the sorted feature groups as a first hierarchical structure; combining the first feature group with a next feature group of the first feature group in the sequenced plurality of feature groups to obtain a next hierarchical structure of the first hierarchical structure; And taking the next hierarchical structure as a new first feature group, and returning to execute the step of m