Local community detection aims at finding a local community from a start node in a network without requiring the global network structure. Similarity-based local community detection algorithms have achieved promising performance, but still suffer from high computational complexity. In this paper, we first design a unified local community detection framework for fusing different node similarity measurement. Based on this framework, we implement eight local community detection algorithms by utilizing different node similarity measurements. We test these algorithms on both synthetic and real-world network datasets. The experimental results show that the local community detection algorithms implemented in our framework are better at detecting local community compared with related algorithms. That means the performance of discovering local community would be largely improved by using good node similarity measurements. This work provides a novel view to evaluate similarity measurements, which can be further applied to link prediction, recommendation system and so on.
CITATION STYLE
Liu, J., Wang, D., Zhao, W., Feng, S., & Zhang, Y. (2017). A unified framework of lightweight local community detection for different node similarity measurement. In Communications in Computer and Information Science (Vol. 774, pp. 283–295). Springer Verlag. https://doi.org/10.1007/978-981-10-6805-8_23
Mendeley helps you to discover research relevant for your work.