Pirá (Pirá: A Bilingual Portuguese-English Dataset for Question-Answering about the Ocean)

A large set of questions and answers about the ocean and the Brazilian coast both in Portuguese and English. Pirá is a crowdsourced question answering (QA) dataset on the ocean and the Brazilian coast designed for reading comprehension.

The dataset contains 2261 QA sets, as well as the texts associated with them. Each QA set contains at least four elements: a question in Portuguese and in English, and an answer in Portuguese and in English. Around 90% of the QA sets also contain human evaluations.

Pirá is, to the best of our knowledge, the first QA dataset with supporting texts in Portuguese, and, perhaps more importantly, the first bilingual QA dataset that includes Portuguese as one of its languages. Pirá is also the first QA dataset in Portuguese with unanswerable questions so as to allow the study of answer triggering. Finally, it is the first QA dataset that tackles scientific knowledge about the ocean, climate change, and marine biodiversity.