NLP Grammar Research

Introduction

This repository is a collection of approaches for applying English grammar correction to a given text.

Approaches:

1. Rule-based approachq

With the rule based approach we developed our own set of rules to correct the grammar of a given text. The rules are based on the English grammar rules and are implemented in Python.

Libraries used:

• NLTK
• Spacy

2. Machine Learning approach

With the machine learning approach we used a pre-trained model Roberta to correct the grammar of a given text. The model is trained on a large dataset of English text and is able to correct the grammar of a given text. The code is an official implementation of the papers "GECToR – Grammatical Error Correction: Tag, Not Rewrite" (BEA-20) and "Text Simplification by Tagging" (BEA-21)