Mining hidden communities in social networks using KD-Tree and improved KD-Tree

摘要

A Social network structure contains several nodes which are connected based on the relationships. Network community mining methods are used discover all hidden communities in distributed social networks based on some criteria. Several algorithms have been developed to solve the hidden community mining problem. In a given network the links between the nodes are opaque, but few nodes are thin. Finding hidden communities in such a large network is always difficult. The existing algorithms like LM (Community mining based on Local Mixing properties) algorithm gives new methods for characterizing network communities via introducing a stochastic process on networks. And it analyzes the network dynamics based on the large deviation theory concept. Through our literature survey we identified few problems in the existing methods. The actual numbers of communities are identified using the recursive bisection methods. Stopping criterion values are predefined. It does not increase communication performance and network partitioning became complex. To overcome the above mentioned problems proposed two algorithms. First we proposed community bipartition method by using KD-Tree. The stopping criterion is calculated automatically. In that we found few limitations, so we proposed an Improved KD tree algorithm. It improves the effectiveness and scalability. In this paper we analyzed both the algorithms that are LM with KD tree and Improved KD tree.