DARC-IT: A DAtaset for reading comprehension in Italian

In this paper, we present DARC-IT, a new reading comprehension dataset for the Italian language aimed at identifying 'question-worthy' sentences, i.e. sentences in a text which contain information that is worth asking a question about. The purpose of the corpus is twofold: to investigate the linguistic profile of question-worthy sentences and to support the development of automatic question generation systems.