Rough Sets and Learning by Unification

Issue title: New Frontiers in Scientific Discovery – Commemorating the Life and Work of Zdzislaw Pawlak

Article type: Research Article

Authors: Buszkowski, Wojciech

Affiliations: Adam Mickiewicz University in Poznań, University of Warmia and Mazury in Olsztyn, Poland. E-mail: [email protected]

Abstract: We apply basic notions of the theory of rough sets, due to Pawlak [30, 31], to explicate certain properties of unification-based learning algorithms for categorial grammars, introduced in [6, 11] and further developed in e.g. [18, 19, 24, 25, 26, 9, 28, 14, 15, 3]. The outcomes of these algorithms can be used to compute both the lower and the upper approximation of the searched language with respect to the given, finite sample. We show that these methods are also appropriate for other kinds of formal grammars, e.g. context-free grammars and context-sensitive grammars.

Journal: Fundamenta Informaticae, vol. 75, no. 1-4, pp. 107-121, 2007