AI Bot for Academic Schedules using Rasa

Abstract

Chatbots or virtual assistants are being used by industries all over the world, they can reduce human intervention and improve efficiency. These days smart-assistants such as Amazon Alexa and Google Assistant help users get quick access to most generic queries within seconds, but when it comes to students and their everyday queries, these assistants fall short in answering the queries they have related to their academic schedules i.e., timetable queries, online classes links, syllabus queries, test dates, etc. The motive behind building this chatbot is to help students get quick and accurate responses to their schedule and syllabus-related queries, this is especially beneficial for students who are taking online classes due to the COVID-19 pandemic and cannot talk to their peers face to face. This chatbot was developed with the Rasa, it is a framework for developing contextual AI assistants and chatbots. Rasa enables the use of components in the NLU pipeline to customize the intent classification, entity extraction, and response selection. This paper goes through the pipeline customizations that were necessary to process the schedule-specific queries from students. It also goes over the stories and custom actions used to generate responses once the intent and entities are extracted. Telegram was used to deploy the chatbot onto the real world to enable students to talk to this chatbot from the comfort of their smartphones.