Frameworks and Algorithms for Regional Knowledge
30.09.2010 16:30 Uhr


Informatik-Hauptgebäude (50.34), HS -101 (UG),
Am Fasanengarten 5, 76131 Karlsruhe


Prof. Dr. Christoph Eick, University of Houston, Texas, USA


Plakat [PDF]

Frameworks and Algorithms for Regional Knowledge

It has been pointed out in the literature that most relationships in spatial data sets are geographically regional, rather than global. Consequently, regional knowledge plays a key role for analyzing and understanding spatial datasets. A generic region discovery framework is presented that is augmented with families of parameterized measures of interestingness that are capable to capture what domain experts are interested in.

The proposed framework views region discovery as a clustering problem and the goal of region discovery is to find a set of spatial clusters (regions) that maximize an externally given reward-based fitness function. Representative-based, grid-based and agglomerative clustering algorithms for region discovery in spatial datasets will be introduced and compared. Experimental results of applying the proposed framework and algorithms to identifying hotspots and collocation patterns in spatial datasets, to learn regional regression functions and to model regional differences in behavior of web users are presented. Finally, the framework is generalized to analyze related datasets.

Short Bio of the Speaker: Christoph F. Eick is an Associate Professor in the Department of Computer Science at the University of Houston and the Director of the UH Data Mining and Machine Learning Group. He received a PhD degree from the University of Karlsruhe in 1984. His research interests include data mining, machine learning, evolutionary computing, geo-graphical information systems, knowledge-based systems, and artificial intelligence. He published more than 100 papers in these areas. He serves on the program committee of the IEEE Data Mining Conference and other data mining and machine learning conferences.