Automatic Question Generation (AQG) is a process by which computer systems generate natural-language questions (interrogative-form sentences).
The canonical case for AQG is the generation of questions from text (such as news, reading passages, and books). A variety of different methods have been developed for text-based scenarios. Another type of AQG is the generation of questions from ontologies, taxonomies, and knowledge-bases. The uses of AQG are also varied. The most often cited purpose is building AQG systems for education, e.g. for test generation, as well as systems for personalized and self-paced learning. Such uses are related to the assessment of reading comprehension and subject-matter domain knowledge, including in Intelligent Tutoring systems. But other use cases may arise, e.g. for entertainment, for games, for dialog systems, etc. While typical uses of AQG are human-centered, a different kind of AQG has recently emerged – AQG as a data augmentation method for training automatic Question Answering (QA) systems.
This Research Topic is intended to provide an overview of current research on AQG, as well as use cases and lessons from implemented systems. We encourage both in-depth technical work and broad interdisciplinary approaches, and invite researchers from a broad range of fields – natural language, machine learning, databases and ontologies, cognitive science, education and language learning, from academia and from industrial settings – to present their achievements and share their perspectives on this rapidly developing area.