Artificial Intelligent Decoding of Rare Words in Natural Language Translation Using Lexical Level Context and Fuzzy Semantic Reasoning

Abstract

Artificial intelligence based Machine Translation is a Natural Language Processing, attaining significant attention to fully computerize the system that can decode basis content into the target languages. The proposed method is a Neural Machine Translation, end-to-end system, with the sentence context for converting German sentences to English sentences. The Transformer based Machine Translation is used for translation integrated with fuzzy semantic representation for handling the occurrence of rare words in the sentence level context. The FSR organization uncommon phrases collectively and constitute sentence degree context as Latent Topic Representation. WMT En-De bilingual parallel corpus is used for translation handling the Out of Vocabulary words using clustering of tags. In the present approach there's a mismatch of translation however the proposed device is extra advanced because of the sentence context inclusion. The model performance is enhanced with hyper parameter optimization obtaining a BLEU score with a better translation of source to target language. Finally minimizing the TER score to attain a better translation rate