Multi-strategy English Translation System Based on DTW Algorithm

Abstract

AbstractWith the continuous development of computer technology and the continuous in-depth study of English translation theory, an English Translation System (ETS) based on the combination of translation and computer has emerged. ETS can automatically perform English-Chinese translation through search. As a commonly used pattern recognition method, DTW algorithm is gradually applied to the field of English information retrieval and translation. Based on the advantages of the current translation methods, this paper designs a multi-strategy ETS based on the DTW algorithm. Based on the traditional DTW algorithm, this paper proposes a two-dimensional dynamic time warping algorithm (2D-DTW), which extends the distance-calculated samples from a spatial one-dimensional sequence to a spatial two-dimensional matrix. This article puts forward the goal of multi-strategy ETS, and plans the composition of the entire ETS realization. Finally, a translation quality evaluation experiment was carried out. The experimental results showed that the average BLEU of the system was 0.6681; the averages of Rec and Pre were 91.39% and 91.13%, respectively. This shows that ETS has better phrase recognition performance. While ensuring that the phrase translation is obtained, it also guarantees the quality of the translation. KeywordsDynamic time warping algorithmEnglish translationPhrase translation acquisitionMultiple strategies