An Automatic Translation Evaluation System Based on Semantic Similarities and Fuzzy Neartude

Most current translation assignments and tests are evaluated in manual work instead of machine, as automatic translation assessment has been considered to be very difficult. In this paper, a system was proposed to evaluate the quality of translation based on semantic similarities and fuzzy neartude. The experiment results are promising and indicate that the system can identify the semantic characteristics and yield scores that correlate with human judgments of translation quality. Our future work is to improve the evaluation strategies and introduce more linguistic resources.