A Semantic Similarity Measure Method Combining Node-based and Distance-based Approaches in Taxonomy
유광청, 유종화, 이병욱 · 한국인터넷정보학회 학술발표대회 논문집 · 2008
Semantic similarity measures play an important role in information retrieval and information integration. There were two categories of approaches for evaluating semantic similarity based on lexical taxonomy: Distance-based approach and Node-based (Information Content Based) approach, however we find both of them Jacks more or less. In the Distance-based approach any two adjacent nodes have a same distance value but in fact the similarity value are not necessarily equal, and the Node-based approach does not differentiate the similarity values of any pair of concepts in a sub-hierarchy. In this paper we present a new measure of semantic similarity between words and concepts in a lexical taxonomy structure. It combines a Node-based method and a distance-based method, and we prove that it works better than both of them.