一种基于语料的词汇语义相似度认知算法

A Cognition Measurement of Word Similarity Based on a Specific Subject Corpus

词汇的语义相似度计算是信息检索、自然语言处理、推荐系统等技术的基础。事实上,词汇可能因其语境或语料的不同,语义的相似度千差万别。论文通过提取词汇的上下文语境特征,构建了一种基于特定语料的词汇的语义相似度计算模型。实验结果表明,该算法有较好的准确率和较强的领域敏感性,取得了令人满意的结果。

Word similarity measurement is the basis for techniques of information retrieval,natural language processing,rec. ommender systems,and so on. Actually,word similarity may be entirely different because of its context or subject. The paper intro. duces a cognition measurement of word similarity based on a specific subject corpus through word's context feature. An experimental result indicates that the measurement method can achieve good precision rate and field sensitivity.