Research on the Link Prediction Model of Dynamic Multiplex Social Network Based on Improved Graph Representation Learning

Tianyu Xia,Yijun Gu,Dechun Yin

doi:10.1109/access.2020.3046526

Tianyu Xia, Yijun Gu + Show 1 more

Open Access

https://doi.org/10.1109/access.2020.3046526

Copy DOI

Abstract

In the natural and social systems of the real world, various network can be seen everywhere. The world where people live can be seen as a combination of network with different dimensions. Link prediction formalizes the interaction behavior between people. Traditional link prediction methods mainly study the user behavior of static social network. This article studied the dynamic graph representation learning so as to put forward an improved link prediction model in dynamic social network. Besides, the interactions in the real world can be multiple, links at different moments may have different meanings. The proposed model firstly solved the problem of link prediction on multiple kinds of edges. The whole embedding of each node is separated into two parts, basic embedding and edge embedding. Then the proposed model selected time slices for dynamic social network to get the graph embeddings in different snapshots. What's more, the t+1 time step embedding vector was used to validate t time step prediction effect and the proposed model performed better than traditional graph representation learning methods.

Highlights

As a complex network data analysis tool, link prediction can be used to process and mine various types of network information, such as assisting scientists in conducting biological protein structure analysis experiments to discover the interactions between different amino acids [1], helping merchants in product recommendation systems to recommend products and services to potential customers [2], assisting data engineers in data processing to retain hidden links and clean up false links [3]
This article introduces an improved link prediction model, which is to take snapshots at equal intervals according to the time sequence, and conducts graph representation learning based on these network snapshots to obtain the vector representation of nodes at different time steps
THE CONSTRUCTION OF IMPROVED LINK PREDICTION MODEL DGATNE we present a model, DGATNE, which is capable of learning representations in dynamic multiplex social network

Summary

INTRODUCTION

As a complex network data analysis tool, link prediction can be used to process and mine various types of network information, such as assisting scientists in conducting biological protein structure analysis experiments to discover the interactions between different amino acids [1], helping merchants in product recommendation systems to recommend products and services to potential customers [2], assisting data engineers in data processing to retain hidden links and clean up false links [3]. T. Xia et al.: Research on the Link Prediction Model of Dynamic Multiplex Social Network Based at the meeting and establish contact. This article introduces an improved link prediction model, which is to take snapshots at equal intervals according to the time sequence, and conducts graph representation learning based on these network snapshots to obtain the vector representation of nodes at different time steps. Classic link prediction methods are mainly based on the similarity of social network structures. These methods do not need to consider the attribute information of the node itself, but only need to consider the local network structure information where the node is located. The similarity between the two nodes depends on the number of resources passed

LINK PRECTION METHOD IN DYNAMIC SOCIAL NETWORK

Compute the self-attention coefficient

Findings

CONCLUSION