Querying the fragments of english

Abstract

Controlled languages are fragments of natural languages stripped clean of lexical, structural and semantic ambiguity. They have been proposed as a means for providing natural language front-ends to access structured knowledge sources, given that they compositionally and deterministically translate into the (logic-based) formalisms such back-end systems support. An important issue that arises in this context is the semantic data complexity of accessing such information (i.e., the computational complexity of querying measured w.r.t. the number of instances declared in the back-end knowledge base or database). In this paper we study the semantic data complexity of a distinguished family of context-free controlled fragments, viz., Pratt and Third's fragments of English. In doing so, we pinpoint those fragments for which the reasoning problems are tractable (in PTime) or intractable (NP-hard or coNP-hard). © 2011 Springer-Verlag.