Three Workshops in Data Analysis and Mining

Rutgers University New Brunswick, New Brunswick NJ

Investigators

Abstract

Theoretical and algorithmic approaches to data analysis have played a central role in the development of modern methods for handling data. The massive amounts of data gathered in important present-day applications ranging from the Internet to homeland security to astronomy and medicine have dramatically changed the requirements for algorithms and provide ample motivation for a great deal of new theoretical development. Traditional data analysis tools are incapable of handling the sheer size and complexity of these gigantic data sets. The project will run three workshops on topics related to the size and complexity of modern data sets: "Data Quality, Data Cleaning and Treatment of Noisy Data" "Compact Representation of Data" "Geographical Information Systems: Terrain Representation and Analysis Algorithms" Intellectual Merit: The Data Quality workshop seeks to bring together experts from different research disciplines to initiate a comprehensive technical discussion on data quality, data cleaning and treatment of noisy data. The objectives of the workshop are to develop quantitative ways of measuring the quality of data and to search for an integrated, domain-independent approach to data cleaning in the situation where the sheer size of data collections prevents one from manually scrubbing or even monitoring them. The central topic of the Compact Representation workshop revolves around the following question: Can a large and potentially complicated data set be replaced by a short sketch such that a certain class of queries on the original data can be answered (perhaps approximately) by performing computations on the short sketch? The workshop will explore metric embeddings of sampling spaces with complex structure into spaces that are amenable to simpler analysis; space-time tradeoffs for data structures; and algorithms for analysis of data in the streaming model. Geographical Information Systems (GIS) are called on to store, combine, display, and analyze disparate types of spatially-referenced data---locations of telephone poles, streets, rivers, mountains, buildings, soil layers, pollution levels, voting districts---for applications ranging from natural resource management to urban planning. The goal of the proposed workshop is to address the challenges in modern GIS by focusing on algorithmic issues encountered in such systems. The workshops will identify areas for research and focus on future research challenges. Broader Impact: All workshops will seek a broad impact through a mix of theoretical researchers and practitioners and by helping to stimulate interdisciplinary collaborations. Theoretical progress on the research topics in these workshops will have impact on a wide variety of areas such as health care, homeland security, environmental management, disaster management, marketing, bioinformatics, psychology, sociology, chemometrics, astronomy, and credit card fraud detection. A major theme of the project is integration of research and education. It is addressed through participation of graduate students in the workshops, interaction between students and DIMACS visitors working in related areas, and tutorial programs organized in each workshop.