Mining Hidden Networks and Sequential Patterns

Real world entities are linked together in sophisticated ways via physical or virtual, homogeneous or heterogeneous links. Natural and social activities, such as network traffic, scientific, biological and engineering processes, generate huge amounts of data in which hidden sequential patterns can be discovered. Mining hidden networks and sequential patterns may detect unnoticed social networks, terrorist or crime rings, core clusters, trends, intrusions, and alarming incidents for critical applications.

Jiawei Han of the Department of Computer Science at UIUC and Michael Welge at NCSA are working on effective, efficient, and scalable data mining methods, building on joint research involving developing efficient sequential pattern and graph mining algorithms. The goal is to construct a user-oriented hidden network and sequential pattern miner for effective and scalable pattern discovery and application development. The work will integrate social network analysis, linkage analysis, sequential and graph pattern mining methods.

The project will lead to the development of a series of effective and efficient data mining methods. These methods should be highly scalable, constraint-based, user-friendly, and very responsive. They are designed for homeland security, social network analysis, and directed marketing. The research results will be published in reputed international conferences and journals. The newly developed approaches will be demonstrated in the D2K data mining framework.