ECAI-2002 Conference Paper

[full paper]

Object Identity as Search Bias for Pattern Spaces

Francesca A. Lisi, Stefano Ferilli, Nicola Fanizzi

In the context of frequent pattern discovery, we present a generality relation, called theta-oi-subsumption, which is based on the assumption of Object Identity in spaces of patterns to be intended as existentially quantified conjunctive formulae. The resulting generality order <=oi seems appropriate for organizing efficiently the space of Datalog patterns over structured domains. Indeed we prove the existence of ideal refinement operators for <=oi-ordered spaces and the monotonicity of <=oi with respect to pattern support. Features of such spaces are illustrated by means of an example of frequent pattern discovery in spatial data.

Keywords: Data Mining and Knowledge Discovery, Knowledge Representation, Search

Citation: Francesca A. Lisi, Stefano Ferilli, Nicola Fanizzi: Object Identity as Search Bias for Pattern Spaces. In F. van Harmelen (ed.): ECAI2002, Proceedings of the 15th European Conference on Artificial Intelligence, IOS Press, Amsterdam, 2002, pp.375-379.