基于改进LPA的重叠社团挖掘算法研究

With people's connection becoming closer, there are many large scale communities in social networks. Detecting communities is very important to reveal the characters of networks. The nodes in one community may belong to another community at the same time, making communities overlapped. How to detect overlapping communities remains a complex problem and we need to do deeper research about it. The Label Propagation Algorithm (LPA) is a simpler and faster community detection algorithm compared with traditional algorithms based on modularity. Although improved LPA could detect overlapping community. it's too slow with constrained detecting abilities. To address these problems, we proposed an algorithm based on improved LPA, which is called Attenuated Overlapping Label Propagation Algorithm (AOLPA). Experiments show that this new algorithm is better than traditional LPA in efficiency and overlapping detecting abilities.