Bisonbot
ChatBot using Open AI and chainlit
This is the part of my Senior Project: under-developing project so I currently do not have any demo for this project but will be adding soon.
The BisonBot project is a revolutionary AI-powered system designed to enhance the user experience on Howard University's website. Faced with the challenge of navigating a complex site structure filled with excessive information, users often experience frustration and difficulty in finding the necessary content. BisonBot addresses these issues by providing a user-friendly, accessible chatbot that guides users through the website, offering personalized assistance 24/7. By leveraging Natural Language Processing and GitHub OAuth for secure logins, BisonBot simplifies the search for information, ensuring that users can efficiently navigate the website and find what they need without hassle.
Key features of BisonBot include the ability to log in with GitHub, initiate conversations on any university-related topic, and receive context-based responses. The system is designed to be intuitive, allowing users to explore popular topics, view chat histories, and even listen to responses, enhancing the overall user interaction. Additional features planned for future updates include third-party login integration, multilingual support, and an embedded chat icon on every page, which will further improve accessibility and user satisfaction.
Behind the scenes, BisonBot operates on a robust backend developed in Python and integrated with the OpenAI API, supported by a frontend that uses Chainlit for an engaging user interface. The system processes data through a series of steps that include data gathering from Howard webpages, converting this data into vector embeddings, and using these embeddings to generate accurate responses to user queries. This comprehensive approach ensures that BisonBot not only meets but exceeds the needs of Howard University's diverse community by providing a responsive and highly functional virtual assistant.
Tech used
- Chainlit
- LangChain to create pipeline for Retrieval-Augmented Generation application
- ChromaDB to store embedding data