A Random Walk Approach to Accurate Condensed Co-Change Graph

Recently, there have been many advances in the research underlying condensed co-change graph, but most research has focused on the discovery of the graph feature. There still few works have done with consideration that the mass unchanged artifacts profoundly effect on the clustering. This paper presents a method for constructing more accurate condensed co-change graph. By using random walk, all the unchanged artifacts can be merged completely on the condition that the changed artifacts greater than 50%. Analysis proves this approach is eligible.