Personalized Web Page Ranking Based Graph Convolutional Network for Community Detection in Attribute Networks

Identifying key information from complex networks is of great practical significance for discovering the community structure. How to effectively use node connections in network and attribute a major challenge current detection problem networks. A graph convolutional based on personalized web page ranking algorithm proposed this paper. First, uses strong characteristics convolution integrating topology information, combines with decouple prediction process propagation model. In addition, improved density peak method used sample local structure center as training set Finally, k-means cluster vector representation results get division. The experimental 7 datasets 13 comparison algorithms show that has obvious improvement detection.