Chatbot inteligente para soporte TI basado en rasa y Telegram con NLP

Abstract

The growing dependence of higher education institutions on information technology has significantly increased the demand for technical support services. At the Salesian Polytechnic University, this situation manifests itself in a high number of requests related to connectivity, access to systems, and use of academic platforms, which are handled manually, resulting in long response times, overburdened technical staff, and a decline in the perception of service quality. Given this problem, the objective of the research was to develop and implement an intelligent chatbot for IT support, based on the Rasa framework and integrated with the Telegram platform, which would automate the handling of frequent requests and improve the user experience. The methodology applied corresponds to applied research with a mixed approach. Interviews with support staff, surveys of students and teachers, and analysis of historical incident records were used to identify recurring patterns. Subsequently, the chatbot's conversational architecture was designed by defining intents, entities, and dialogue flows, training the Natural Language Processing model with contextualized data. The results obtained showed a significant reduction in response times for frequent queries, a decrease in the operational workload of support staff, and greater user satisfaction. The system demonstrated its ability to accurately interpret requests and provide consistent and timely responses.