The Machine Translation Model
Ziyuan Zhao · Advances in Social Science, Education and Humanities Research/Advances in social science, education and humanities research · 2022
Machine Translation was invented in the late-20 th century, when the first IBM model could automatically translate Russian sentences into English.Proceeding to the 21 st century, linguists have invented new types of Machine Translation models and improved on them by altering and adding parts.Of all the various types of Machine Translation models, this paper will mainly focus on Statistical Machine Translation models, which put into play the statistical data with their calculation of the probabilities of the words and phrases, and Neural Machine Translation models, which are built by the stacking and connecting of neural layers through encoders and decoders, by comparing the benefits and flaws of the types within the specific Machine Translation models.A specific translation will be mentioned using the Neural Machine Translation models, revealing the flaws within its translation.From the flaws of Neural Machine Translation models, this paper will also examine attempts to improve them by solving existing problems.This paper will build a basic understanding of machine translation models and possibly inspire future experiments and extensions to improve the translation models both mentioned and not mentioned in this paper.