This study is based on the fact that machine translation evaluation is closely related to the sentence- level semantic equivalence. According to what writer says this study will answer this question whether there is any correlation between performance on the semantic equivalence classification task and performance of underlying evaluation technique on the task of MT evaluation? In this paper some methods or standard methods are used in order to build organizers to predict semantic equivalence. The most important method which is new and novel is the PER. This model leverages parts of speech information. Which is related to the word matches and non- word matches through the sentence. The most significant part, now, is the process of dealing with synonyms. The writer said the meaning of a sentence is conveyed by a synonymous word in its paraphrase. In fact, the writer covered these cases in order to get help from them for proving the result of the study. A pilot study is also done for the better results. Finally, it was indicated that it is possible to drive the features that can be used to determine whether similar sentences are paraphrases of each other from methods currently being used to automatically evaluate machine translation systems.

To sum up, this study done in line with sentence equivalence in machine translation. It is based on the similarity between the assessment in machine translation and the task of sentence- level semantic equivalence. The results supported the study completely. The technique gives a substantial improvement in analysis or paraphrase classification accuracy over all of the other models used in the experiments.