Foundations of Inductive Databases for Data Mining

1 January, 2006 to 31 December, 2009


  • Hendrik Blockeel
  • Maurice Bruynooghe

The project aims at developing the foundations of an inductive database. The goals are threefold: 1. develop an adequate theory for data mining in the context of inductive databases, in the form of a representation language to store the data as well as the inductively acquired knowledge. This representation language will be based on fragments of first order logic. 2. apply this theory in practice by studying how existing data mining techniques fit into it, such as association rules, classification, clustering, and several others. 3. develop efficient implementations of the proposed models. An important research question here is to what degree existing optimization techniques in data mining can be generalized.