Description
DiffQG: Generating Questions to Summarize Factual Changes DiffQG is a dataset that consists of questions that summarize the factual change between two versions of the same Wikipedia passage. This requires detecting whether a factual change occurred, given a noun phrase, and if so, generating a question that summarizes it.
DiffQG is a small, evaluation-only dataset consisting of expert annotations. More details can be found in our EACL paper.