Mongolian-English, English-Mongolian independent Neural Machine Translation System

For more than a decade, PMT and SMT models have dominated the field of machine translation, and neural machine translation has emerged as a new paradigm for machine translation. The latest neural machine translation not only performs better than systems that consider the structure of ordinary words and sentences, but is also able to find complex relationships between source and target words. Neural machine translation provides a simpler modeling mechanism that makes it easier to use in practice and science. Neural machine translation no longer requires concepts such as word rank, which is a key component of a system that takes into account word and sentence structure. While this simplicity can be seen as an advantage, on the other hand, the lack of careful wording is a loss of control over translation. Systems that take into account word and sentence structure generate translations that consist of word sequences in the training data. On the other hand, neural machine translation is more flexible for translation that does not exactly match the training data. This provides more opportunities for such models, but frees the translation from predefined constraints. Lacking a specific word connection can make it difficult to link the target words you create to the source word. The widespread use of neural machine translation in translation systems has the advantage of allowing users to translate certain terms and translate untrained data to a certain extent, but in some cases often results in distorted sentence structure. This paper aims to address issues such as neural machine translation control, more accurate translation of unrecognized data, correct sentence structure and grammar boundaries, and the creation of independent machine translation system.