Mining and Monitoring Evolving Data

Venkatesh Ganti and Raghu Ramakrishnan

Data mining algorithms have been the focus of much recent research. The initial spurt of research on data mining algorithms typically considered static datasets. In practice, the input data to a data mining process resides in a large data warehouse whose data is kept up-to-date through periodic or occasional insertion and deletion of sets of tuples. Consequently, several issues that arise in a dynamically evolving database have recently begun to receive widespread attention. In this article, we survey research on two important issues: (1) exploiting the systematic data evolution for efficiently maintaining data mining models, and (2) monitoring changes in data characteristics. We classify research addressing these two problems based on a few distinguishing characteristics, and then briefly discuss all techniques captured by this classification.

Keywords: Data mining, Dynamic databases, Evolving data, Monitoring data characteristics.