一种计算组合词汇语义相似度的方法  

A METHOD FOR COMPUTING SEMANTIC SIMILARITY OF COMBINATION WORDS

作　　者：吴学军[1] 何丰[1] 

机构地区：[1]北方民族大学计算机科学与工程学院,宁夏银川750021

出　　处：《计算机应用与软件》2010年第8期215-218,共4页Computer Applications and Software

摘　　要：对语义相似度的研究,已产生了许多计算模型和计算方法。这些模型和方法大多集中于简单词汇之间的相似度计算,很少有涉及两个或多个词汇组合的相似度计算。在本体匹配与服务发现的过程中,与本体有关的概念不可避免的会出现由两个或多个词汇组合成的组合词汇。在WordNet相似度计算基础上,综合考虑了词汇间的各类相似度,提出了一种计算组合词汇的语义相似度的方法。实验表明,该方法能有效提升相似度计算的有效性和准确性。There are a lot of computing models and methods with regard to the researches about semantic similarity. Most of these models and methods focus on computing similarity among simple words, but few involve the similarity computation of two or more lexical combinations. In the process of ontology matching and service discovery, the concepts which related to ontology might form combination words inevitably, which consists of two or more words. Based on the WordNet similarity computation,in this paper we consider comprehensively all kinds of simi- larity among words, and propose a method to compute the semantic similarity of combination words. Experimental results show that this method can improve validity and accuracy of similarity computation efficiently.

关 键 词：本体 组合词汇 语义相似度 WORDNET 

分 类 号：TP391.12[自动化与计算机技术—计算机应用技术]