A novel overlapping community detection algorithm in smart city combining interest topic and local density

Abstract It is very important to use ontology user portraits to describe the semantic structure interests of users and to study the similarity relations among user portraits for discovering overlapping interest communities in the process of building smart cities. The user interest’s hierarchical characteristics can produce different similarity relationships, forming interest clusters. The current paper proposes an overlapping community identification method based on hierarchical characteristics of user interest and local density is proposed, which can effectively promote the construction of smart cities. Firstly, a user hierarchical interest model based on ontology knowledge base was constructed to determine the users’ multi-granularity topic similarity. Then, a heterogeneous hypergraph was constructed by using multi-granularity topic similarity and user following similarity to describe the interest network. Using the peak mechanism of interest density, the community detection method was applied to identify the communities of interest. The real performance of the algorithm on multiple networks was verified by experiments. Experimental results showed the superiority of the presented algorithm over the classical overlapping community identification approach in terms of precision and recall rate.