Network alignment method based on double-layer graph attention neural network

Abstract

The invention provides a network alignment method based on a double-layer graph attention neural network. The network alignment method comprises two stages of network embedding representation and embedding vector space alignment. In a network embedding representation stage, a double-layer graph attention neural network is provided to carry out network representation learning so as to extract an embedding vector of a user in a social network; in an embedded vector space alignment stage, a classification model is constructed by utilizing the obtained social network user node embedding vectors and a part of known anchor link sets to predict anchor links between different social networks, and a bidirectional embedding vector space alignment strategy is proposed to meet one-to-one matching constraints of user entities between different social networks. Through the above setting, the method can effectively capture different influence weights of the user in the social network and between theneighbor users and the features, thereby learning the precise representation of the user in the social network, and improving the prediction accuracy of the anchor link between different social networks.

Description

technical field [0001] The invention relates to the technical fields of data mining and machine learning, in particular to a network alignment method based on a double-layer graph attention neural network. Background technique [0002] With the rapid development of the Internet and mobile devices, online social networks have become an indispensable and popular platform for people to share and exchange information. Since different social networking platforms provide different services, a person usually registers accounts on multiple social networking platforms at the same time to meet their different needs. These users shared by different social network platforms naturally form anchor links connecting different social networks, which promotes information interaction between different social networks. Mining information interactions across multiple social domains can be effectively applied to various downstream social network applications such as cross-domain link prediction,...

AI Technical Summary

Problems solved by technology

Although these methods try to model the user's behavior in the social network from multiple aspects such as the user's social structure and profile information, they ignore the difference in the

Images