Vyakranly : Hindi Grammar & Spelling Errors Detection and Correction System

Abstract

The growing demand for automation tools for Hindi over the past few years has led NLP experts to start working towards tasks that facilitate research and development for Hindi Language Processing. Researchers have been increasingly putting efforts into building models to perform essential NLP tasks like spell correction, grammar correction, summarizing, and so on. Compared to the plethora of tools and data available for English, Hindi is a relatively new area in which not much work has been done so far. Therefore, researchers need to build frameworks and technologies that support the Hindi language to perform such complex tasks effectively.However, with limited data and tools available for Hindi, performing Grammatical Error Correction (GEC) for Hindi may come across as a challenge. Therefore, we propose Vyakranly (व्याक्रणली), a Hindi Translation and Grammatical Error Detection Toolkit for the Indic language Hindi. Objectives of Vyakranly (व्याक्रणली) are Hindi Text Spelling Error Detection and correction, Hindi Sentence Grammar Error Detection and correction, English to Hindi and Hindi to English text translation. Highlights of our work are Hindi spelling detection as well as correction and grammar error detection. It is difficult to find such past work for the Hindi language due to their relative lack of digitized content and complex morphology, compared to English.