Ontologies and text: can NLG bridge the gap?
Richard Power · International Conference on Natural Language Generation · 2010
Ontologies are akin to technical documents in that they describe domain knowledge, but they express this content very differently, in formal languages like OWL designed for use by machines, not people. During the last decade, interest has grown in the task of mapping from OWL to controlled fragments of natural language, thus providing a niche for NLG in which we are at last agreed on the formal specification of the input. The aim of the talk is to compare ontologies and their textual counterparts (e.g., technical dictionaries, encyclopedias) at several linguistic levels. After looking at their usage (pragmatic level), we will consider terminology (roughly, word level), statements (sentence level), and groups of related statements (discourse level). The question is whether we can find similar levels in the organisation of OWL ontologies, thus allowing a mapping from ontologies to texts that can be exploited by NLG systems.