PathRAG Court

Full Project Report here

Running the project: 🚀

• Since Pathway is meant to be deployed in a containerized manner, and single-machine deployments can easily be achieved using Docker - we opted to make the whole app be able to run in a docker container. The dependencies are automatically installed into the container using the requirements.txt file.
• Therefore to run the project, you need to have Docker installed in your environment.
• First, go into the project directory, and setup a .env file (have to make the file) containing all the API Keys required to run the LLMs and fetching tools.:

GOOGLE_API_KEY= <your_key_here>
GROQ_API_KEY= <your_key_here>
HUGGINGFACE_API_KEY= <your_key_here>
SERPER_API_KEY= <your_key_here>
KANOON_API_KEY= <your_key_here>
// Other API keys for any LLM you would like to use

• Then, run the following command:

docker build -t pathwaytest .

pathwaytest can be replaced with any name you want to give to the image. Then, run the container with the following command:

docker run -it -p 8000:8000 --rm --env-file .env pathwaytest

Use the same name as above, this will expose the backend FastAPI server at port 8000.

---

Architecture Diagram: 🏛️

Problem Solved: 🎯

1. Preparing for Legal Cases

   • Problem: Preparing for a court case involves analyzing previous cases, predicting arguments, and anticipating the course of proceedings, which can be overwhelming.
   • Solution: By simulating court proceedings based on user-provided case descriptions and files, the system helps users practice and prepare for real-world legal scenarios, providing insights into potential arguments, counterarguments, and outcomes.

2. Educational Tool for Legal Understanding

   • Problem: Many individuals lack exposure to how courts operate, leading to confusion and anxiety about legal proceedings.
   • Solution: The system provides an educational platform where users can interactively learn about court processes, understand legal principles, and explore how different factors influence verdicts.

3. Accessible Legal Expertise

   • Problem: Access to professional legal expertise is often expensive and not always feasible for all individuals, especially for preliminary case evaluations.
   • Solution: By simulating courtroom interactions with intelligent agents, the application provides users with a preliminary understanding of their case, enabling informed decisions without immediate legal consultation.

References

1. Indian Penal Code PDF: link
2. PDF Parsing reference: Multimodal RAG for PDFs with Text, Images, and Charts
3. Vector store reference: Data Indexing