A context-aware chatbot designed for UCI students, offering quick access to course details and tailored responses to their queries. Featuring a robust course information database, Langchain retriever, and prompt template, the chatbot delivers efficient and personalized assistance.
Watch the Live Demo: https://youtu.be/VWcLdm3qz3I?si=FpXSdaONgbQ9MKQU
Tech used: Python, Streamlit, LangChain, FAISS, HuggingFace Transformers, BeautifulSoup
CourseBuddy is built as a Retrieval-Augmented Generation (RAG) system to provide accurate and contextually relevant responses about UCI courses. The development process involves several key components:
- HTML Parsing: Utilized BeautifulSoup to scrape course information from UCI's official course catalog.
- Text Conversion: Converted HTML content into plain text files for processing.
- Embeddings: Generated vector embeddings using HuggingFace's
WhereIsAI/UAE-Large-V1model. - Vector Database: Stored embeddings in a FAISS vector store for efficient similarity search.
- Language Model: Integrated HuggingFace's
Mistral-7B-Instruct-v0.2model via API. - Prompt Template: Designed custom prompts to guide the model's responses based on retrieved context.
- Retrieval Chain: Implemented a retrieval chain using LangChain to fetch relevant documents and generate answers.
- Streamlit App: Developed an interactive UI with Streamlit, allowing users to input queries and receive responses in a chat-like format.
- Session Management: Maintained conversation history using Streamlit's session state for a seamless user experience.
- Reduced Hallucinations: By grounding responses in retrieved context, the chatbot minimizes the risk of generating inaccurate information.
- Efficient Retrieval: FAISS enables fast and scalable similarity searches over the embedded course data.
- Modular Design: The system's components are modular, facilitating easy updates and maintenance.
Developing CourseBuddy provided insights into:
- RAG Systems: Understanding the integration of retrieval mechanisms with generative models to enhance response accuracy.
- User Experience: Designing intuitive interfaces that effectively convey AI-generated information to users.