Educational Chatbot to Support Question Answering on Slack

Educational chatbots unlock new possibilities to support teaching activities. In recent years, the progress made in natural language processing and understanding allowed the development of virtual assistants that understand questions written as unstructured data. They generate a proper answer based on knowledge bases. This work proposes an accurate solution easy to test in real world scenarios like university courses. In fact, the use of Slack and RASA technologies reduces the difficulties to train and install the educational chatbot on already existing digital workplaces. We evaluates the effectiveness of our solution reaching 0.93 accuracy in intent recognition building a stacked neural network.