Hidden community attribute acquisition method and system based on attention graph neural network

Abstract

The invention discloses a hidden community attribute acquisition method based on an attention graph neural network. The method comprises the following steps: learning all vocabularies in a user social media data vocabulary library through a word vector model Word2vec network to obtain embedded representation vectors of all the vocabularies; performing normalization weighting on embedding representation vectors of user social media data vocabularies, and obtaining embedding representation of a target user based on a target embedding layer of a forward full-connection network; based on the user social network and the social activity information, generating embedded representations of neighbor users of the target user, and calculating a weight social matrix according to the embedded representations of the neighbor users; and training a social popularity weighted attention graph neural network according to the weight social matrix, and generating a hidden community attribute classification result of the target user by using the attention graph neural network and the embedded representation of the target user. The invention also provides a corresponding hidden community attribute acquisition system based on the attention graph neural network.

Description

technical field [0001] The invention belongs to the technical field of attribute mining, and more specifically relates to a method and system for acquiring hidden community attributes based on an attention graph neural network. Background technique [0002] User research in cyberspace is one of the important tasks in the field of Internet personalized recommendation. With the continuous expansion of social network scale, user attribute information in cyberspace has the characteristics of sparsity, fragmentation and heterogeneity, which leads to It is very difficult to obtain some hidden community attribute information of target users, which makes it difficult to carry out the next analysis and recommendation work. How to mine potential or hidden community attribute information of users in social network platforms through effective technical means is crucial to analyze the potential preferences of target users on the content to be recommended. [0003] For the cognition anal...

AI Technical Summary

Problems solved by technology

[0003] For the cognitive analysis of user attributes, the attribute features used in existing research are mostly lightweight features, which contain insufficient information, and the exploration of the target user’s attribute embedding representation is insufficient, and there is often still a large amount of information to be mined.

Secondly, the classification of target user attributes produced by existing research is still too simple, often targeting relatively simple classification targets such as the target user’s gender and educational background. There is still a large space for exploration in content behavior representation methods

Images