Sinhala Grammar Suggestion Tool Using NLP Hybrid Approaches

Abstract

According to the surveys of the author, an automated grammar checking tool for the Sinhala language is a shortage so far. But seeing how difficult it was to understand the factors of the sentences and how extensive work had to be done to scale the system to introduce new grammatical characteristics, it was realized that the rule-based approach was not so suitable for this scenario. Therefore, a neural machine translation model was created to overcome these issues. This project is focused on correcting the free word order nature and multiplex morphology of the Sinhala language. This proposed solution will help to rectify probable Sinhala grammar errors and also give respective correct suggestions for each error. After developing this application, it was trained with predetermined grammatical errors that are related to verbs in Sinhala language. And its performance was exceeding expectations and was highly promising. Also it is hoped that its application will be convenient, given the fact that it’s really easy to use.