Abstract: Taking consideration of complexity and dynamic of complex networks, a definition of local modularity was proposed, and an algorithm for communication structure mining based on local information detection was given, with the criterion, i, e. the best community is the node group whose local modularity is the largest. Then a method of multi-granularity community structure mining was proposed, which provides new ideas to observe structure characters of complex networks from various angles. Final experiments verify its efficiency and feasibility.

Key words: Complex networks, Community structure mining, Local information detection, Multi granularity