Motivating Ontology-Driven Information Extraction

Abstract. Ontologies can provide Information Extraction Systems (IES) with much needed domain and task knowledge. Yet, it has to be analysed to what extent and in which form ontologies can be utilised to enhance the overall performance of an IES. Further, the use of ontologies requires an accurate management of the same. The most important issue with this respect is to keep the ontology up-to-date, because the domain it represents will change inevitably. In this paper we motivate the use of ontologies within IES, especially for automating the extraction-rule generation process, which currently is the main obstacle to portable and scalable IESs.