Demographic attribute inference method based on cross-platform user social multimedia behavior

Abstract

The invention discloses a population attribute inference method based on cross-platform user social multimedia behavior. Wherein, the method includes extracting text features and picture features for the social multimedia behaviors of the users on the platforms to obtain the user features of the platforms; using the demographic attributes of the users in the training set and the user features as supervisory information , performing dual projection matrix learning in multiple spaces, and constraining the social multimedia behavior of each platform of the user, so as to determine the dual projection matrix of each platform and share stable user information; according to the user characteristics Performing projection transformation on the dual projection matrix, and inferring the population attribute based on the shared and stable user information obtained as a result of the projection transformation. The embodiment of the present invention solves the contradiction between the observed dynamics of social multimedia behavior and relatively stable demographic attributes, and improves the accuracy of the user's demographic attribute inference results.

Description

technical field [0001] The embodiment of the present invention relates to the technical field of data processing, in particular to a population attribute inference method based on cross-platform user social multimedia behavior. Background technique [0002] With the popularity of social media (social media), more and more people participate in various online social networks, and publish and share a large amount of multimedia information. In order to achieve efficient digital information management and personalized social media services, user modeling through social multimedia behavior is becoming more and more urgent and important. User modeling includes many aspects, from demographic attributes (such as: age, gender, marital status, and occupation, etc.), personal interests (such as: politics, technology, music, and sports, etc.) to social network status, mobility patterns, and consumption patterns. and emotional tendencies. Among them, the demographic attribute records t...

AI Technical Summary

Problems solved by technology

However, a very important issue has been neglected and under-studied at present: the contradiction between the observed dynamics of social multimedia behavior and the relatively stable population properties

In the context of such demographic attribute inference, considering that demographic attributes including gender, age, marital status, and occupation are static or remain unchanged for a long period of time, the method of dynamic interest modeling cannot be directly applied.

Images