Survey on Hinglish to English Translation and Classification Techniques

Abstract

Code-mixing is the process of using many languages in one sentence and has a widespread occurrence in multilingual communities. It is particularly prevalent in texts on social media. Due to the widespread usage of social networking sites, a substantial amount of unstructured text is produced. Hinglish, i.e. code-mixed Hindi and English, is a frequent occurrence in everyday language use in India. Hence, a translation process is required to help monolingual users and to aid in the comprehension of language processing models. In this paper, we study the effective techniques for classification and translation tasks and also find gaps and challenges in the current research domain. After comparing a few existing methodologies for machine translation, a framework which showed an improvement in task of translation over the previous methods is proposed.