摘要： 提出一种计算WordNet中概念间语义相似度的算法，该算法同时考虑概念的信息内容(IC)以及2个概念在WordNet is_a关系分类树中的距离信息，由此提高算法性能。给出一种计算概念IC值的新方法，通过考虑概念的子节点数及概念所处WordNet分类树中的深度，使计算结果更精确。与其他5种语义相似度算法的比较结果表明，该算法能够求得更准确的相似度。

关键词: 信息内容, WordNet本体, 语义相似度, 子节点, 分类树

Abstract: A new algorithm of calculating semantic similarity of concepts in WordNet based on Information Content(IC) is presented. The model considers the IC value of concepts and their positions in the is_a taxonomy tree in WordNet, which improves the performance of the model efficiently. Furthermore, a new method of calculating IC value of concept is given. The method considers the number of child node of concept and its depth in the taxonomy tree of WordNet, which makes the IC value more accurate.

Key words: Information Content(IC), WordNet ontology, semantic similarity, child node, taxonomy tree

中图分类号: 

• TP312