Summer 2007 REU Project

      Data Mining, Search, and Retrieval Systems

Student Participants Ralph Abbey (North Carolina State University) Jeremy Diepenbrock (Washington University) Dexin Zhou (Bard College) Advisors Carl D. Meyer (Primary Faculty Advisor, NC State) Shaina Race (Graduate Student Advisor, NC State) Amy Langville (Secondary Faculty Advisor, College of Charleston) Project Description The project focused on mathematical and computer technology involved in building various kinds of data mining, search, and retrieval systems. The search and retrieval methods ranged from classical latent semantic indexing systems through Google's pagerank technology. The data mining components revolved around a variety of data clustering schemes. Conclusions & Results New clustering techniques involving gapped PDDP and extended Fiedler algorithms were developed. In addition, a new concept of cluster aggregation was formulated that appears to outperform many standard clustering algorithms. Details are given in the final report. Presentations Poster presentation given at the Undergraduate Research Symposium, McKimmon Center, NC State University, August 2, 2007. Information retrieval and web search tutorial (pdf), July 17, 2007. Data Clustering and Mining (.ppt) Mid-Hudson Mathematics Conference (presented by Dexin Zhou) Bard College, October 20, 2007 Final Report This is the final project report (pdf) that was submitted on August 1, 2007.