Automatic Scoring for Translations Based on Language Models

摘要

With the development of English education, translation scoring has gradually become a time-consuming and energy-consuming task, and it is difficult to ensure objectivity because of the subjective factors in manual correcting. Due to the similarity between the quality evaluation of responses generated by the dialogue system and the translation results submitted by students, we selected two metrics of dialogue to automatically score the translations, which are applied in a case study. The experiments show that the hybrid scores of two metrics are close to human scores. In conclusion, the method is feasible to apply the evaluation metrics of dialogue systems to translation scoring, and it can provide an improvement idea for the automatic scoring of translations in the future.