Active Semi-supervised Community Detection Algorithm with Label Propagation

Abstract

Community detection is the fundamental problem in the analysis and understanding of complex networks, which has attracted a lot of attention in the last decade. Active learning aims to achieve high accuracy using as few labeled data as possible. However, so far as we know, active learning has not been applied to detect community to improve the performance of discovering community structure of complex networks. In this paper, we propose a community detection algorithm called active semi-supervised community detection algorithm with label propagation. Firstly, we transform a given complex network into a weighted network, select some informative nodes using the weighted shortest path method, and label those nodes for community detection. Secondly, we utilize the labeled nodes to expand the labeled nodes set by propagating the labels of the labeled nodes according to an adaptive threshold. Thirdly, we deal with the rest of unlabeled nodes. Finally, we demonstrate our community detection algorithm with three real networks and one synthetic network. Experimental results show that our active semi-supervised method achieves a better performance compared with some other community detection algorithms.