Overlapping Network Community Detection Through Label Propagation Algorithm with K-Shell Aggregation

Overlapping community detection in large-scale networks poses significant challenges due to the high complexity of traditional algorithms. The Label Propagation Algorithm (LPA) stands out for its efficiency in large networks but struggles with identifying overlapping communities accurately. The proposed LPAKSA (Label Propagation Algorithm with K-Shell Aggregation) addresses these limitations by integrating an information aggregation method for assessing node significance, improving node similarity calculations, and refining label updating strategies. Tested on both real and synthetic networks, LPAKSA demonstrates marked improvements in accuracy and stability for detecting overlapping communities and maintaining linear time complexity.