Experiments on generating questions about facts
This paper presents an approach to the problem of factual Question Generation. Factual questions are questions whose answers are specific facts: who?, what?, where?, when?. We enhanced a simple attribute-value (XML) language and its interpretation engine with context-sensitive primitives and added a linguistic layer deep enough for the overall system to score well on user satisfiability and the 'linguistically well-founded' criteria used to measure up language generation systems. Experiments with open-domain question generation on TREC-like data validate our claims and approach. © Springer-Verlag Berlin Heidelberg 2007.
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Rus, V., Cai, Z., & Graesser, A. (2007). Experiments on generating questions about facts. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 4394 LNCS, 444-455. https://doi.org/10.1007/978-3-540-70939-8_39