AI Chatbots for the International Mobility Unit of Politecnico di Torino

Conversational AI chatbots play an important role across both institutions and industries, revolutionizing the way organizations interact with users. Their ability to provide instant, personalized responses has made them essential tools for improving customer service, streamlining processes, and enhancing user engagement. The goal of this master's thesis is to explore the development of an AI-powered chatbot designed to facilitate the process of answering frequently asked questions (FAQs) and handling ticket inquiries for prospective students at the Politecnico di Torino (PoliTO) International Mobility Unit. Leveraging the Rasa platform, this project implements both rule-based and generative AI approaches to enhance the efficiency and accuracy of responses. The primary objective is to streamline the communication process for students who are not yet admitted but are interested in applying for undergraduate or graduate programs. The rule-based approach, which utilizes predefined patterns and responses, ensures consistent and reliable information delivery. Concurrently, the generative AI approach, based on models like T5, aims to provide more flexible and context-aware interactions. This research includes pre-processing diverse datasets, including FAQ data and ticket logs, each with unique characteristics, to train and fine-tune the chatbot models. Performance metrics such as ROUGE and BERTScore are used to evaluate and compare the effectiveness of the different approaches. The findings demonstrate that while the rule-based model excels in precision and reliability, the generative AI model significantly improves the handling of complex and varied queries. This work not only contributes to the existing body of knowledge on AI chatbots but also provides a practical solution to enhance the student experience at PoliTO by automating and optimizing the response process for prospective applicants.