Research of calculating sentential similarity in translation memory
翻译记忆中语句相似度计算方法的研究

Several commonly-used calculating methods of the sentential similarity, one of the key technologies of Translation Memory, were analyzed. According to the requirements of Engllsh-Chlnese translation memory system, a new method based on sentential form and semantics was proposed. The method stored the sentences on the basis of sentential forms, and obtained the sentential similarity with combining the semantic similarity and the weight of the component of the sentences. It can display the degree of semantic and structure between two English sentences commendably.