Finding Overlapping Communities by Increasing the Determinacy of SLPA in Complex Networks

Community detection algorithms help to analyze the real structure of networks. The advantages of overlapping community detection algorithms are evident in discovering high-quality communities. SLPA is a classical algorithm of overlapping community detection, which is updated by random sequence in the phase of label updated. At the same time, it randomly selects one of the labels to update when a node receives multiple labels that appear most frequently in the phase of label propagation. Aiming at the uncertainty of randomness in the algorithm, an improved algorithm of SLPA based on nodes PageRank values and nodes similarity index is proposed in this paper. A large number of experiments have been carried out in benchmark data sets and real data sets. The results show that the proposed algorithm has excellent adaptability and robustness