Improvised Neural Machine Translation Model for Hinglish to English

Code-mixed language is a crucial part of communication in multilingual communities, and with the recent increase in internet usage, its use has also grown. Social media communication has evolved over the past few years due to the shift from desktops and laptops to smartphones. Social media posts are more casual and concise, and people do not feel the need to write fully formed and grammatically correct sentences in any particular language. One language that has gained prominence due to this phenomenon is Hinglish, a hybrid language that combines Hindi and English. It is widely spoken in India and is used by people who are fluent in both languages. Hinglish has emerged as a solution to bridge the gap between those who speak only Hindi and those who are fully bilingual, providing a contemporary and region-specific way of communication that is easily accessible to the general public. However, the unstructured hybrid texts in Hinglish need to be translated into a standard machine-understandable form to aid in data analysis. Proposed research has explored the effectiveness of Neural Machine Translation (NMT) architectures, including Sequence-to-Sequence (STS) and Attention (ATT) models, for this task. NMT is a cutting-edge method of automated language translation that uses artificial neural networks to understand the mapping between two languages. NMT has surpassed the performance of earlier approaches like Statistical Machine Translation (SMT) and has experienced significant growth over the past few years. It is a powerful way to translate from Hinglish to English and has demonstrated promising outcomes. The use of NMT has the potential to revolutionize the field of machine translation and enable seamless communication between people speaking different languages.