Comparative Analysis of Machine Translation (MT) and Computer Aided Translation (CAT)

Abstract

Since the emergence of MT, its fast translation speed and declining cost have been favored by the whole society. In order to make MT more useful, computational linguists have been committed to improving the accuracy of MT. By discussing the advantages and disadvantages of computer-aided translation and the advantages and disadvantages of MT, this paper compares the changes of translation accuracy, recall rate and word reasoning ability under different sentence lengths under different translation modes of MT, computer-aided translation and human translation through the test of machine frame model algorithm. The results show that in the length range of 0–10, the translation accuracy of the three translation models has reached more than 85%, which shows that the additional non top-level information of the encoder has an obvious gain for the model in translating shorter sentences; In the length range of 10–20, the translation accuracy does not decrease significantly; In the range of 30–40, the accuracy of sentence translation begins to decline significantly, especially in MT. With the increase of sentence length, the accuracy decreases to 70% in the range of 40–50; The decline of computer-aided translation is not obvious, and the accuracy is 79% in the length of 40–50; The accuracy of human translation is the most stable, maintained at more than 85%; There is little difference in the recall rate, which has reached more than 85 points, but the score of word reasoning is slightly different. The highest score is MT, which has reached 88 points, which fully proves the effectiveness of MT. Based on the above experimental test results, it can be seen that MT and computer-aided translation have their own advantages, and there is little difference in the overall level. For the translation accuracy, computer-aided translation combines human translation on the basis of MT, and computer-aided translation has more advantages than MT.