Julian Szymański1 and Wlodzislaw Duch2,3,
1Department of Electronic, Telecommunication and Informatics, Gdansk University of Technology, Gdansk, Poland.
2 School of Computer Engineering, Nanyang Technological University, Singapore,
and 3Department of Informatics, Nicolaus Copernicus University,
Grudziadzka 5, 87-100 Torun, Poland.
Understanding information contained in texts and symbols requires extensive background knowledge. This knowledge is stored in semantic memory. Recently manually created ontologies and dictionaries became very popular but still do not contain sufficient information for many applications, such as semantic search or automatic annotation of concepts. Architecture facilitating knowledge acquisition for common sense semantic memories is presented and an algorithm for playing 20 questions word game used for further verification and acquisition of new knowledge. This approach is then applied to the WordNet database for gathering new and correcting existing knowledge.
Keywords: knowledge acquisition, commonsense knowledge, ontologies, semantic memory, Wordnet, text understanding, semantic Internet, NLP.
Reference: The Sixth International Conference on Information and Management Sciences (IMS2007), July 1-6, 2007, California Polytechnic State University, CA, pp. 342-348
Preprint for comments in PDF, 302 KB.
BACK to the publications of W. Duch.
BACK to the on-line publications of the Department of Informatics, NCU.