Chatbot interactivo basado en voz para responder a preguntas frecuentes acerca de la Universidad Técnica Particular de Loj

Abstract

Abstract: In this thesis, a voice-based interactive chatbot, represented by a 3D avatar, was developed to answer frequently asked questions about the Universidad Tecnica Particular de Loja (UTPL). The solution implemented in this thesis adopts a monolithic architecture deployed using Docker Compose, which integrates the frontend, backend, and an Ollama-based natural language processing service. These components are orchestrated in containers to ensure isolation and scalability. The system integrates various technologies on the frontend, such as Vue.js, Three.js, and TypeScript, along with Python, LangChain, and ChromaDB on the backend, providing a fluid and accurate experience. The adapted agile methodology (SCRUM) enabled a structured implementation in planning, development, and testing phases, ensuring quality and efficiency. The incorporation of speech recognition and synthesis (Speech-to-Text and Text-to-Speech) and a 3D avatar optimized user interaction, making it more intuitive and engaging. This innovative tool responds to the needs of the university community, improving access to institutional information with low operating costs and the ability to operate offline, positioning itself as a modern, accessible, and high-impact solution.