Institut National des Sciences Appliquees de Lyon

We consider the challenge of supporting KDD processes by means of queries. We introduce a research perspective, the so-called inductive database framework [4, 1, 3, 2]. An inductive database is a database that in addition to data contains intensionnaly dened generalizations about the data. Specifying and implement-ing inductive databases is a long-term perspective. However, more or less special-ized data mining query languages have been proposed, e.g., MSQL [7] or MINE RULE [5]. It is interesting to discuss whether they are good candidates for in-ductive database querying. Furthermore, the optimization of inductive queries is clearly related to constraint-based mining (see e.g., [6]). Considering the asso-ciation rule mining task, we can gure out what should be done in a near future in order to implement inductive databases for that simple but useful kind of pattern. However, applying the framework and studying the robustness of the identi ed concepts for other data mining tasks remain open. Considering these issues is the main goal of the European project cInQ IST-2000-26469 (May 2001-May 2004), funded by the Future and Emerging Technologies arm of the IS