Using WordNet for Word Sense Disambiguation to Support Concept Map Construction1

The construction of a concept map consists of enumerating a list of concepts and —a more difficult task— determining the linking phrases that should connect the concepts to form meaningful propositions. Appropriate word selection, both for concepts and linking phrases, is key for an accurate knowledge representation of the user’s understanding of the domain. We present an algorithm that uses WordNet to disambiguate the sense of a word from a concept map, using the map itself to provide its context. Results of preliminary experimental evaluations of the algorithm are presented. We propose to use the algorithm to (a) enhance the “understanding” of the concept map by modules in the CmapTools software that aide the user during map construction, and (b) sort the meanings of a word selected from a concept map according to their relevance within the map when the user navigates through WordNet’s hierarchies searching for more appropriate terms.