Query answering over ontologies using controlled natural languages

Abstract

In this thesis we define and study the expressive power and the data complexity of a certain number controlled languages for ontology-based data access systems (OBDASs), in which data stored in relational databases is queried through an ontology (i.e., a so-called conceptual or intensional) middle layer that can be modeled by description logics. Controlled languages and controlled language interfaces have been proposed as a means of enhancing the usability of interfaces to information systems, and, in particular to OBDASs. Controlled languages are subsets of natural languages (such as English) with a limited vocabulary and syntax, designed to avoid the ambiguity inherent to arbitrary natural language utterances. Their design however typically ignores issues like their complexity or scalability to the information queried, which as we show can vary greatly depending on their coverage of basic English syntax.