NOCD: a new overlapping community detection algorithm based on improved KNN

In social networks, the community detection algorithm is very important for understanding the structures and the functions of these networks. A lot of researches have been done on the overlapping community detection algorithms as the overlapping is a significant feature of such networks. However, though many algorithms have been introduced to detect overlapping communities, the detection of the overlapping community is still a challenging task. In fact, the traditional static methods which partitioned the network structure could not efficiently obtain the latest community structure. The problems of high computational complexity and low identification accuracy need to be solved. To address these issues, in this paper, we propose a New Overlapping Community Detection algorithm based on improved KNN (called NOCD), which can timely adjust the community structure based on different network changes, and ultimately obtains the results of the community partitions with a high degree of Q module. To deal with the weighted social networks, NOCD adopts similarity instead of distance to evaluate the network. The experimental results show that the proposed NOCD algorithm compared with the COPRA, the CPM, the DeCom, the PLPA, and the AI-LPA algorithms can effectively improve the detection accuracy, the efficiency of parallel computing, and reduce the time complexity.