15.A community detection framework based on the feature transfer and incremental community detection

Abstract:

A community detection framework based on feature transfer and incremental community detection is proposed to address dynamic community detection problems.A multi-phase incremental community detection algorithm for dynamic networks provides a higher-precision initial community partitioning. Valuable information from the previous time slice is extracted to the next one using feature transfer, combined with the improved SPEA2 algorithm, aiming to maximize clustering accuracy within each time slice and minimize differences in clustering results between consecutive time slices for multi-objective optimization.Experiments on synthetic and real datasets demonstrate superior performance in performance metrics compared to mainstream community detection algorithms.

Keywords: incremental community detection;dynamic networks;feature transfer;multi-objective optimization;time series