Towards automatically retrieving discoveries and generating ontologies

Abstract

For the web to become intelligent, machines needs to be able to extract the nature and semantics of various concepts and the relationships between them. Most approaches focus on methods involving manually teaching the machine about different entities, their properties manually constructing an ontology. This paper discusses an approach where the necessary metadata is extracted automatically from Wikipedia, the online encyclopedia. This metadata is then used to compare documents allowing them to be clustered together so that similar documents can be identified allowing alternative knowledge to be discovered. The results show that an ontology indicating the relationships between types of documents can be automatically identified and also alternative knowledge can be discovered.