Despite exponential accumulation of electronic texts in the Internet age, effective search and reuse is notrnavailable for most texts due to the lack of association among them. Terms and the semantic relationshiprnbetween terms have great potential in organizing and sharing textual information. We develop an agent-basedrnapproach for discovering term relationship in texts collected from the Web. The goal of the paper is torninvestigate which of the five generic relationships in text, including synonym, hypernym, hyponym, holonymrnand meronym, can be effectively discovered with the proposed approach.
展开▼