JJE: Applying Graph Flow Theory to Wikipedia

Data in the world, and more specifically on the Internet is growing to massive sizes. In order to make this information more useful, it must first be more accessible. The INEX Initiative competition is aimed with the goal of identifying and comparing methodologies for categorizing information into clusters. The competition will be run on 60 gigabytes of data from Wikipedia, with the ultimate goal of accurately categorizing and clustering in order to reduce search time through this data. In this work, we construct a clustering algorithm based of the link structure of a subset of underlying pages. The resulting webgraph is pruned using a max flow min cut algorithm[10, 2, 8] which is initially seeded using different heuristics. We compare search space reduction results and construct a visualization of the clustered documents. We were able to generate clusters on the INEX data set as well as visualization of clustered data on several different datasets.