Read more
This book provides an overview of the fundamentals of Automatic Question Generation (AQG) for computational linguistics researchers, test developers, and educators. The author presents a variety of AQG system architectures, including generating questions from syntactic analyses, semantic resources, neural architectures, ontologies and knowledge graphs, and large language models. The advantages and pitfalls of a variety of AQG evaluation methods, including multi-aspect ratings by human experts, end-users, as well as crowd-sourcing and automatic evaluation techniques are discussed. The book also provides a roadmap of options for AQG targeted orientation, content selection, and focusing decisions. Machine learning opportunities for training systems to generate questions based on human-generated examples are also explored. This book offers greater depth and breadth than previous surveys of AQG. Readers will gain a comprehensive knowledge of current research, examples of applications of AQG, and inspiration for future directions for innovation and application.
About the author
Michael Flor is a senior research scientist at the ETS Research Institute, the research arm of the Educational Testing Service, in Princeton, New Jersey. He specializes in language technology for educational applications. At ETS, Dr. Flor has led, and co-led R&D projects focused on text complexity and readability, question generation, automated spelling correction, analysis of figurative language, automated scoring of essays and short answers, assessment of collaborative problem solving, sentiment analysis, and on psycholinguistic aspects of the lexicon. He serves as an academic editor for the PLOS ONE journal and as a guest associate editor for the journal Frontiers in AI. He is a long-time member of SIGEDU and has been a committee member for multiple international conferences and workshops in the areas of NLP and language learning.
Summary
This book provides an overview of the fundamentals of Automatic Question Generation (AQG) for computational linguistics researchers, test developers, and educators. The author presents a variety of AQG system architectures, including generating questions from syntactic analyses, semantic resources, neural architectures, ontologies and knowledge graphs, and large language models. The advantages and pitfalls of a variety of AQG evaluation methods, including multi-aspect ratings by human experts, end-users, as well as crowd-sourcing and automatic evaluation techniques are discussed. The book also provides a roadmap of options for AQG targeted orientation, content selection, and focusing decisions. Machine learning opportunities for training systems to generate questions based on human-generated examples are also explored. This book offers greater depth and breadth than previous surveys of AQG. Readers will gain a comprehensive knowledge of current research, examples of applications of AQG, and inspiration for future directions for innovation and application.
