First Elements on Knowledge Discovery guided by Domain Knowledge (KDDK)

Jean Lieber (1) , Amedeo Napoli (1) , Laszlo Szathmary (1) , Yannick Toussaint (1)

1 ORPAILLEUR - Knowledge representation, reasonning

In this paper, we present research trends carried out in the Orpailleur team at Loria, showing how knowledge discovery and knowledge processing may be combined. The knowledge discovery in databases process (KDD) consists in processing a huge volume of data for extracting significant and reusable knowledge units. From a knowledge representation perspective, the KDD process may take advantage of domain knowledge embedded in ontologies relative to the domain of data, leading to the notion of ''knowledge discovery guided by domain knowledge'' or KDDK. The KDDK process is based on the classification process (and its multiple forms), e.g. for modeling, representing, reasoning, and discovering. Some applications are detailed, showing how KDDK can be instantiated in an application domain. Finally, an architecture of an integrated KDDK system is proposed and discussed.

Mots clés: knowledge discovery; classification; formal concept analysis; data mining

Domaines: Intelligence artificielle [cs.AI]

Jean Lieber, Amedeo Napoli, Laszlo Szathmary, Yannick Toussaint. First Elements on Knowledge Discovery guided by Domain Knowledge (KDDK). Fourth International Conference on Concept Lattices and Applications - CLA 2006, Oct 2006, Hammamet, Tunisia. pp.22--41. ⟨hal-00608015⟩