Analysis through NLP techniques of didactic interactions on social channels. The thesis describe the creation of a Chatbot on a communication channel called Slack, able to understand the recurrent questions made from an university course students and if the question has already been answered by a professor in the past, it proposes an answer. If the question is a new one, it gives the opportunity to create an open question to be answered by a professor and later on it makes it available for future interactions. Also the research analyze and compare some different NLP (Natural Language Programming) pre-trained models (SpaCy, BERT, ConveRT) used in the edu-bot in learning process.