Material Detail

Knowledge Discovery from Evolving Data

Knowledge Discovery from Evolving Data

This video was recorded at European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases (ECML PKDD), Antwerp 2008. Data mining has traditionally concentrated on the analysis of a static world, in which data instances are collected, stored and analyzed to derive models and take decisions according to them. More recent research on stream mining has put forward the need to deal with data that cannot be collected and stored statically but must be analyzed on the fly. At the same time, the need to store, maintain, query and update models derived from the data has been recognized and advocated [LT08]. However, these are only two aspects of the dynamic world that must be analyzed with data mining: The world is changing and so do the accumulating data and, ultimately, the models derived from them. The challenges for Knowledge Discovery in a changing world have two forms: (a) adapting the patterns to the changes in the population and (b) capturing, understanding and highlighting the changes. In this tutorial, we discuss the topics associated with data mining for changing environments and elaborate on research advances in this area. Relevant research comes among else from the fields of incremental mining, stream mining, temporal mining and change detection. Since this is a very wide field, we concentrate on the second challenge, the understanding of change, and we organize research contributions in this context.


  • User Rating
  • Comments
  • Learning Exercises
  • Bookmark Collections
  • Course ePortfolios
  • Accessibility Info

More about this material


Log in to participate in the discussions or sign up if you are not already a MERLOT member.