ICDS Chatbot (PathFinder Team)

Inspiration

Our inspiration for the ICDS RAG Chatbot came from the need to streamline access to crucial information from the vast resources available within the Penn State Institute for Computational and Data Sciences (ICDS). As data science continues to grow in complexity, we saw an opportunity to create a tool that simplifies and enhances user interactions with large-scale datasets and guides, enabling researchers and professionals to quickly find accurate answers to their queries without navigating through dense technical documentation.

What it does

The ICDS RAG Chatbot is an AI-driven tool that answers questions by retrieving relevant information from the Penn State ICDS resources and generating human-like responses. Using Retrieval-Augmented Generation (RAG) technology, it offers users instant access to data and insights, enhancing efficiency in research and decision-making processes by eliminating the need for manual searches.

How we built it

We built the chatbot using advanced NLP techniques integrated with a Retrieval-Augmented Generation (RAG) model. This combination allows the chatbot to retrieve contextually relevant information from ICDS resources and provide accurate, real-time answers. The backend was developed with Python and integrated with APIs that connect to large data repositories, ensuring up-to-date information is processed. We used Llamaparse, VoyageAI, Cohere, and OpenAI to make this chatbot. Frontend wise, we used typescript, shadcn/ui, tailwindcss, and react.

Challenges we ran into

One of the biggest challenges we faced was ensuring the chatbot retrieved highly accurate and context-specific answers from a vast and complex data set. We had to struggle with parsing particular pdfs and the images and tables that they contained, information that is not as easy to part as text. Additionally, ensuring a smooth user experience and balancing speed with accuracy were ongoing challenges.

Accomplishments that we're proud of

We’re proud of successfully implementing a sophisticated RAG model that significantly improves user interaction with computational data resources. The chatbot’s ability to provide fast, accurate responses and its design that adheres to user-friendly principles is a major accomplishment. Additionally, being able to meet the needs of researchers and professionals at Penn State’s ICDS is a rewarding achievement.

What we learned

Throughout the project, we learned how to efficiently integrate RAG technology with large datasets and how critical user experience is in AI-driven tools. We also deepened our understanding of model optimization and the complexities involved in scaling AI solutions to meet institutional needs.