Many real applications can be modeled using bipartite
graphs, such as users vs. files in a P2P system, traders vs.
stocks in a financial trading system, conferences vs. authors
in a scientific publication network, and so on. The
connections between two entities are the key on mining such
a graph. In particular, we propose algorithms to compute
the neighborhood for each node using random walk with
restarts and graph partitioning; we also propose algorithms
to identify abnormal nodes, using neighborhood information.
We evaluate the quality of neighborhoods based on
semantics of the datasets, and we also measure the performance
of the anomaly detection algorithm with manually
injected anomalies. Both effectiveness and efficiency of
the methods are confirmed by experiments on several real
datasets.