A heterogeneous social network cross-platform associated user account mining method

Abstract

The invention discloses a heterogeneous social network cross-platform associated user account mining method. The invention discloses a heterogeneous social network cross-platform associated user account mining method. The method comprises the steps that for different social network characteristics, respective social network structures are established, the structure of any social network OSNi is represented by a graph G (i) = (V (i), R (i)), V (i) represents a vertex set and is composed of multiple different types of vertex sets in the social network OSNi, and R (i) represents a relation set between vertexes. The heterogeneous social network cross-platform associated user mining method comprises the four steps of construction of a social network structure, extraction of a user interaction element path, migration and embedding of a single social network and cross-social network migration and embedding and associated user account mining. The method has the advantages that cross-platform unified embedding is achieved, and the problem that efficient cross-network platform embedding cannot be achieved in the prior art is solved.

Description

technical field [0001] The invention relates to the field of social networks, in particular to a method for mining cross-platform associated user accounts in heterogeneous social networks. Background technique [0002] Online Social Networks (OSNs) such as Twitter, Sina Weibo and WeChat are constantly integrating into our daily lives, attracting the participation of billions of users around the world. People make friends, express opinions and share information through social networks. information. In recent years, many social networking platforms with different functions and types have appeared, and users can choose to join one or more according to their preferences. For example, after joining twitter, users can publish tweets (tweets) to update status and ideas, Foursquare encourages users to record and share location check-in information, and Instagram brings users the best experience of sharing pictures anytime, anywhere. [0003] The emergence of a variety of social ne...

AI Technical Summary

Problems solved by technology

[0010] Aiming at the problems of narrow application range, sparse and incomplete data and high requirements for training sets in the existing methods, the present invention provides a related user mining method not only applicable to homogeneous social networks but also applicable to cross-platform heterogeneous social networks

First, user interaction meta-paths are extracted from heterogeneous social networks; then, based on transfer learning, triples are defined to embed two users and their interaction meta-paths. On the one hand, it can avoid the difference between different semantic spaces and achieve a unified vector space. Embedding, on the other hand, using meta-paths can avoid the impact of sparse and incomplete data on embedding; finally, using iterative mining can effectively solve the problem of poor mining effect on small-scale training sets

Images