Integration of WordNet into a Topic Selection Routine

WordNet is a lexical database of the English language that seeks to capture the relationships between words as a graph data structure [8]. In the WordNet graph, vertices represent word groupings of synonyms, while edges are labeled relations between either two specific words in a grouping or two synonym groupings. The richness of these relationships allow WordNet to encode information about the English language that is lexical (such as word morphology), syntactic (such as part of speech), and semantic (such as antonym groupings.) The amount of effort required to capture this much information has typically precluded its incorporation into conversational agents. However, with the public availability of WordNet, the WordNet project has assumed the cost of data entry, allowing other projects to freely benefit from its extensive information on the English language. This paper will investigate some of the ways the WordNet database may be integrated into conversational agents, and focus on one particular use of WordNet to extend an existing conversational agent.