Thomas Vestskov Terney, Tine Lassen
In order to facilitate content based information retrieval we need methods for analyzing the semantic structures in the text. In order to facilitate this, we propose a straightforward method for identifying semantic relations between two concepts in an ontology based on supervised machine learning. More specifically we have achieved good results in trying to identify the relation a preposition denotes between two noun phrases, by using their ontological type. However, there is apparently no improvement by including the path from the ontological type to the top of the ontology in the learning process.
Subjects: 1.10 Information Retrieval; 11.2 Ontologies
Submitted: Feb 14, 2006