Metadata made easy - Adding metadata by means of natural language

Abstract

The rapid increase in the amount of information generated highlights the need for better strategies for searching for relevant content. Using Semantic Web technologies we can search using concepts and their properties and relations to each other. To enable the use of Semantic Web technologies documents on the Web need to contain metadata which can describe their content in a way readable by computers. This thesis will describe the development of MaDaME, a prototype system that will help users add this metadata to their documents using only natural language. The tool will utilize WordNet to disambiguate natural language, and as an intermediary to find mappings into formal ontologies. The thesis will examine the feasibility of using the hypernym relation, and the hyponyms of hypernyms, for finding mappings to more general categories when no direct mapping exists.