Finding What Changes for Two Graphs Constructed from Different Time Intervals

Many kinds of graph data including social networks are increasing nowadays. In such a graph, the relationships between vertices are changing day by day. We need to have a data mining method able to extract significant patterns informing us about what changes. From this viewpoint, we propose in this paper an algorithm for change detection over two graphs constructed from two time intervals, based on a notion of modularity and a branch-and-bound search method for changes.