Research On Overlapping Clustering And Attribute Graph Clustering Algorithms

In real life, many objects could be modeled in the form of graph data, and thereare many application values using data mining algorithms on graph data. Clustering isan important method in data mining process. Many researchers have proposed manygraph clustering algorithms to solve this problem. In this paper, the writer proposestwo improved algorithms to solve two problem better, which appear in many problemin clustering graph data.Traditional clustering could find many different clusters, each pairs of whichcould not overlap. This conflicts the real life seriously. For example, every personbelongs to many different communities in social network. Therefore, we should allowsome vertices belong to different communities, and we call this problem "overlappingcommunity detection". Edge-based algorithms could solve this problem and findbetter community structure. In this paper, a novel edge-based algorithms is proposed.In this algorithms, every edge in the graph is converted into multi-edges, and we useoverlapping methods to find edge-based clusters, then convert the result intovertex-based clusters. Tests on real data give good result.Graph can contains not only topology structure, but also node/edge attributes,which is called attribute graph. Attribute graph could describe this world better, anddata mining on attribute graphs could find more accurate or more interesting patterns.In this paper, we could get the weight of every edge, according to node attributes. Wepropose a simple and useful methods to determining the edge weights, then usingweighted graph clustering algorithms to cluster the attribute graph. Attribute weightmatrix could be calculated using EM algorithms. Tests indicate that the methodsproposed can find better clusters than the newest algorithms.