Abstract:
This project presents the design, development, testing, and evaluation of an AI-based semantic chatbot system tailored for tertiary education institutions in Sri Lanka.
The primary objective was to enhance student engagement and support by enabling real-time, context-aware, and personalized responses to academic and administrative inquiries through natural language interaction.
The chatbot system was developed using a modular architecture comprising:
• A web-based user interface built with HTML, CSS, and JavaScript,
• A Flask-based backend server for managing user sessions and API communication,
• Integration with OpenAI’s GPT-4o model for natural language understanding and generation,
• A FAISS vector search engine to perform semantic information retrieval from an institutional knowledge base,
• An SQLite database for persistent user data, chat history, and feedback storage.
The system was designed to operate across all major web browsers and device types, ensuring compatibility with Windows, macOS, Android, and iOS operating systems.
Through comprehensive black box, white box, integration, usability, and exploratory testing, the chatbot demonstrated strong performance in delivering accurate, contextually relevant answers and a smooth, user-centric interaction experience. Key features included multilingual support (informal Singlish handling), chat history retrieval, feedback collection, voice output via the Web Speech API, and dynamic UI enhancements.
Stakeholder evaluations validated the system’s effectiveness, highlighting its potential for broader deployment within educational institutions.
The project successfully showcases how AI-driven semantic chatbots can be integrated into higher education environments to improve information access, communication efficiency, and user satisfaction.
