Sentence Structure and Boundary for Deep Neural Machine Translation Alignment Model

Abstract

AbstractIn terms of social globalization, the world remains more connected than ever due to the widespread use of digital technologies. The translation is a gate for those communication. When we want to understand, study, and express any kind of information prepared in a language other than one's native language, no matter where you are in the world, we will need translation. Due to social media usage, consumers are more likely to get information written by other users in foreign languages, and researchers need to conduct research in many languages and publish the results in one's second languages. Because professional translation requires so much hard work, automated translations, known as machine translations, have played a vital role in helping millions of consumers understand information written in a foreign language. In addition to being used by ordinary users to make every day common translation, it is also possible to help professional translators translate quickly. The modern neural machine translation not only performs better than systems that consider the sentences structure, but is also able to find complex relationships for those translation candidates. It offers a simpler modeling that makes it easier to implement. Neural machine translation no longer requires intermediate steps such as word rank, which is a key component of a system that uses word and sentence structure. While those easiness can be count as a benefit, on the other view, the absence of careful wording is a loss of accordance over translation. On the other corner, neural machine translation is more supple for translation that does not exactly match the training data. The prevalent usage of neural machine translation in translation systems has the benefit of allowing users to translate certain terms and translate untrained data to a positive extent, but in some cases often results in distorted sentence structure and boundary. This paper aims to address issues such as neural machine translation control, more precise translation of unrecognized data, correct sentence structure and boundaries.KeywordsMongolian-English TranslationSentence StructureDeep NMTSentence BoundaryHierarchical Triple Model