Automatic annotation of local data sources for data integration systems

In this article we present CWSD (Combined Word Sense Disambiguation) a method and a software tool for enabling automatic annotation of local structured and semi-structured data sources, with lexical information, in a data integration system. CWSD is based on the exploitation of WordNet Domains, structural knowledge and on the extension of the lexical annotation module of the MOMIS data integration system. The distinguishing feature of the algorithm is its low dependence of a human intervention. Our approach is a valid method to satisfy two important tasks: (1) the source annotation process, i.e. the operation of associating an element of a lexical reference database (WordNet) to all source elements, (2) the discover of mappings among concepts of distributed data sources/ontologies.