P.I.L.L.A.R is a web application designed to simplify the interaction with legal documents and laws, benefiting both lawyers and individuals needing legal insights. Leveraging advanced technologies, P.I.L.L.A.R enhances the accessibility and understanding of legal content through intelligent summarization and semantic search capabilities.
Summary Feature : (40 Page document was used)
- Frontend: Next.js
- Backend: Pytorch, Transformers
- Models: DistilBART, Llama2 LLM
- Search: Elasticsearch
Using P.I.L.L.A.R, users can easily understand and interact with legal documents and laws. Lawyers benefit from accessing various cases for research purposes and case preparation.
-
Summarization of Lease Agreements: Utilizes the DistilBART model to simplify and summarize complex lease agreements, making legal documents more accessible and understandable.
-
Document-Related Q&A Chatbot: Powered by Llama2 LLM, the chatbot assists users in navigating legal documents by answering specific questions related to legal content.
-
Semantic Search: Implements a semantic search feature using Elasticsearch and vector representations, enhancing search accuracy and relevance across a database of 300k legal cases.