Research And Implementation Of The Algorithm In A Complex Network Community Structure Discovery

Complex networks relate to various disciplines. In recent years they have attracted more and more attention and gradually become the focus of lately research. As an important feature of complex networks, community structure plays a decisive role in network analysis. So the discovery of community structure in complex networks is very necessary in the stuty of network structure and function. People have proposed many algorithms for finding community structure in complex networks. This paper studies the classic community structure discovery algorithm, then we propose two new community structure algorithm based on these former algorithms. The main task of this paper is as follows:(1) Complex network cluster-algorithm is an effective method in community structure discovery. Most of the proposed complex network cluster-algorithms are sensitive to the initialization and are easy to get local optimal solution, so they are not applicable for networks whose structure is complex and community is not obvious. In order to solve these problems, We propose a community detection algorithm based on ant colony clustering algorithm(ACCA). Firstly, improved spectral method is used in order to transform nodes into characteristic vector. Secondly, applied ACCA to detect cluster structure. Finally, the best community is selected by computing the function of modularity. We extend the algorithm in order to find community structure in weighted networks. The experimental results show that this algorithm can overcome the shortcomings of proposed cluster-algorithms, and it has highly efficiency and well result.(2) With the development of the times, network size is bigger and bigger in reality. So discover the whole network is time-consuming. On the other hand, local community structure algorithm gradually get the attention of people. Past local community finding algorithms depend on the initial node, and we need know the number of local community nodes. To solve these problems, this paper propose a method based on local network information. In order to find the community which initial node belong to, we use the complex network’s own local network connection feature. From the specified node, search neighbor node continuously, then bring the closest node into community. The extended algorithm can not only find the whole network’s community but also overlapping nodes. The experimental results show that this algorithm overcomes the shortcomings of local community finding algorithm in the past, and it has good effect.