Diversified generation of commonsense reasoning questions

Jianxing Yu, Shiqi Wang, Han Yin, Qi Chen, Wei Liu, Yanghui Rao, Qinliang Su

Abstract: Highlights•We reveal the challenges of yielding commonsense questions with diversity and difficulty control.•We propose a new QG method, with useful reasoning skeletons and expressions to control generation.•The results can meet conditions of commonsense reasoning with expected complexity and diversity.•We conduct extensive experiments to fully evaluate the effectiveness of our approach.•Our commonsense results can facilitate the application of fact-checking.