GraphExplorer: Graph mining for large networks

The primary goal in Phase II is to further develop the graph mining algorithms investigated in Phase I, and to build the technical infrastructure that will enable these algorithms and the visualization interface to be deployed with much larger graphs. There are several key objectives that need to be met in order to build such a system. The foremost issues are to address scalability and visualization. The first we address by investigating ways to optimize our current algorithms. We identified clear paths to do so during our Phase I effort, which we will pursue during Phase II. The second, we address by helping the user identify informative collapsible views or parts of the graph that are worth mining before doing any visualization. We will also focus on expanding the graph mining toolkit to focus on three specific informative types of graph patterns to look for. We will base the enhanced graph mining capabilities on the novel graph statistics we developed in Phase I and pursue the probabilistic view of data which showed the most promise in our evaluations. Phase II will also involve deployment to real applications and we will develop a rapid prototyping environment for graph mining applications to explore the use of GraphExplorer mining tools on commercial data within a commercial product.