Working Groups in Data Analysis and Mining

Rutgers University New Brunswick, New Brunswick NJ

Investigators

Abstract

This project involves three "working groups" of researchers meeting at DIMACS, the Center for Discrete Mathematics and Theoretical Computer Science, to address different aspects of massive data analysis problems in different applied contexts. One group is concerned with "streaming" data analysis and mining. The group deals with computational and analytical methods relevant to problems that arise when decisions must be made in one, initial scan as data "stream" by. The results will have impact on massive data set problems arising from credit card transactions, telephone calling, financial transactions, environmental modeling, and astrophysical experiments. The second group develops and analyzes new models and algorithms for and applications of multidimensional scaling. MDS, a traditional tool of data analysis in marketing, psychology, and other social and behavioral sciences, faces new challenges from the sheer volume of data in today's databases and from a variety of new applications, and the results should be useful in a wide variety of applications in economics, management science, chemistry, and psychology. A third group brings together researchers designing computers to generate scientific conjectures. Specifically, the group deals with conjectures in graph theory and related areas of chemistry that are generated from databases of graph invariants and relations between them and of chemical structures such as of fullerenes. The results should be of wide interest to researchers in graph theory and chemistry but also to researchers dealing with automatic theorem proving and generation. The results of the three working groups will be broadly disseminated to research communities via technical reports, papers, and books in the DIMACS book series.