Towards an automatic Q&A generation for online courses - A pipeline based approach

Abstract

Personalization of online courses is one of the challenges of the 21st century. Although different methodologies for personalization in educational contexts are already existing, there is a bottleneck: personalization by context is always limited to existing learning material; creation of those is a time-consuming task. In this paper we introduce a pipeline to generate questions and valid answers based on educational texts, limited to factual questions for given sentences. We combined NLP technologies with an efficient methodology that is normally used in bioinformatics and adjusted it to generate Q&A-pairs. Instructors can suggest corrections in natural language. Our system generates questions and corresponding answers based on sentences of which 70% make sense.