Customized AI powered website chatbot using retrieval augmented generation (RAG) framework

• Main Author: Zhou, Qiren
• Other Authors: Kong Wai-Kin, Adams
• Format: Final Year Project
• Language: English
• Published: Nanyang Technological University 2024
• Subjects: Computer and Information Science; Chatbot; Generative AI; RAG
• Online Access: https://hdl.handle.net/10356/175274
• Institution: Nanyang Technological University

The integration of Artificial Intelligence (AI) and web technologies has opened up new avenues for enhancing user experience. This project presents the development of an AI-powered website chatbot that leverages the Retrieval Augmented Generation (RAG) framework to provide intelligent and contextualized responses to user queries. The chatbot employs web scraping techniques to extract relevant information from websites, and then utilizes OpenAI's embedding models to convert the content into machine-readable representations. These embeddings are stored in a Supabase database, enabling efficient similarity searches. When a user poses a query, the chatbot generates a standalone question, retrieves the most relevant context from the database, and combines it with the question to produce an accurate and informative response using OpenAI's language models. The frontend is built with React and Next.js, ensuring a responsive and intuitive user interface. The backend is designed for scalability and adaptability, allowing easy integration with different websites. The project includes a comprehensive business plan that outlines a strategic roadmap for commercializing the chatbot technology. The plan proposes an initial phase of offering personalized digital consultancy services, followed by targeting information-intensive websites across various sectors. The long-term goal is to transition into a Software as a Service (SaaS) model, leveraging the scalability and cost-efficiency of cloud computing to achieve rapid market penetration. The business plan also includes pricing strategies, market analysis, and a detailed SWOT analysis to identify strengths, weaknesses, opportunities, and threats. A proof-of-concept prototype tailored for the School of Computer Science and Engineering at Nanyang Technological University demonstrates the chatbot's effectiveness in assisting prospective students and parents with undergraduate admissions queries, showcasing the potential of this AI-driven solution to revolutionize user engagement and information accessibility on websites.