Legal Named Entities Extraction

In the "Legal Named Entities Extraction" Natural Language Processing project, conducted over one month from November to December 2022, I developed two transformer-based deep learning models tailored for entity recognition within Indian legal documents. These models achieved an exceptional F1-score of 90, demonstrating their accuracy and effectiveness.

Leveraging LegalBert for feature embeddings and an LSTM network for entity-level sequence labeling, I implemented sophisticated input sequence encoding and tag-label decoding using the pre-trained transformer, facilitated by SimpleTransformers. The tools at my disposal, including Python, HuggingFace, Keras, TensorFlow, SpaCy, and Pandas, collectively enabled the creation of a robust and highly efficient solution for extracting legal named entities from text data.