Overlapping community detection algorithm based on semantic similarity and label pre-allocation
In order to solve the problems of the label propagation algorithm(COPRA)when dividing communities,the initial label is multiple,the label update is random,and the propagation does not consider the semantics in social networks,we proposed a method OCDSLP to improve the community detection based on semantic similarity and label pre-allocation.Firstly,based on the Word2Vec model,the user's blog post is modeled to measure the user's semantic similarity.Then,the network topology feature or-dering node and the semantic information are used to pre-allocate tags to reduce the initial number of tags.Finally,incorporate node similarity when choosing labels,limit the size of the community,and discover communities with consistent semantics.Experi-mental results have shown that OCDSLP can significantly improve the quality of discovery communities.
semantic informationlabel pre-allocationlabel propagationoverlapping community
万方数据
维普
