A New Translation Strategy Based-on Case-Based Reasoning in Machine Translation System

To improve machine translation system´s translation quality and accurate rate, a case-based reasoning translation strategy was presented. Laying emphasis on the case reasoning part, a structure and semantic similarity algorithm was designed to calculate the sentence similarity. First, a dynamic filter is used to select the translation pattern roughly. Then, based on dynamic programming (DP) algorithm, the structure similarity was decided by the calculation of edit distance. In the process of path recall, an optimization function was defined to select the best matching path. Last, to match the pattern more exactly, the semantic similarity is calculated by the semantic tree structure. The experiment shows that the score of translation quality is 70.11 and the accurate rate can reach 89.4%