Grammar Error Corrector


Project Overview

Grammatical Error Correction is a challenging task in Natural Language Processing (NLP), involving the detection and rectification of errors in text. The complexity arises from diverse vocabularies and language rules. The project aims to address this issue, enhancing writing accuracy and breaking down language barriers, particularly for non-fluent speakers.


Blog

GitHub Repository
Performace Metric

The project adopts the Bilingual Evaluation Understudy (BLEU) score, a widely-used metric in NLP. BLEU assesses the similarity between the machine-generated output and human-generated reference sentences. This quantitative measure provides insights into the effectiveness of the Sequence-to-Sequence models employed in the grammatical error correction process.

Technologies Used

Project Details

Dataset

Two distinct datasets contribute to the training and evaluation of the grammatical error orrection model.

  • Lang-8
  • NUS Social Media Data

    • Approach

    The project adopts a Sequence-to-Sequence model as the foundational approach for grammatical error correction. The task is framed as a Natural Language Processing (NLP) problem, specifically in the domain of Natural Language Processing. The chosen approach involves the application of the Sequence-to-Sequence model, where the deep learning model receives an input sequence (incorrect text) and produces an output sequence (corrected text). The project delves into the Sequence-to-Sequence model's intricacies, employing the Sequence-to-Sequence model with attention mechanisms, such as Loung Attention and Monotonic Attention, to enhance the model's performance.

    Result


    Note: For detailed description check out the blog and Github Repository

    Get in Touch

    Connect with me through the following platforms: