Desarrollo de un Chatbot basado en modelos de lenguaje de gran escala (LLM) para optimizar el proceso de matrícula en el colegio Unión Nacional de Educadores, de la parroquia Chicaña, cantón Yanzatza, provincia de Zamora Chinchipe

Abstract

Abstract: This thesis describes the development of a chatbot based on large-scale language models (LLM) to streamline the enrollment process at the Colegio Unión Nacional de Educadores, in the Yantzaza canton. The main requirements include answering frequently asked questions, guiding users step by step according to their role (student, representative or administrator) and personalizing responses according to the type of user. In addition, the administrator is given the power to update the knowledge base in real time through a friendly interface in Angular and data stored in the PostgreSQL database. Thanks to the integration of the GPT-3.5 Plus model through the OpenAI API, the implementation was carried out quickly and smoothly, avoiding the complexity of developing a language model from scratch. This reduced costs and demonstrated the versatility of the pre-trained models available on the Internet. The chatbot was able to respond accurately, reducing the administrative workload and improving the end-user experience. A usability survey applied to students and teachers reflected results of Very Good or Excellent in the experience of using the application. In short, this project marks a significant step towards the modernization and automation of educational processes and lays the foundations for future administrative applications.