CREPE: Open-Domain Question Answering with False Presuppositions
This work addresses a practical issue for information-seeking users by providing a benchmark to improve QA systems in handling false presuppositions, though it is incremental as it builds on existing QA frameworks.
The authors tackled the problem of open-domain question answering when questions contain false presuppositions, which are common in real-world information-seeking but not addressed by existing datasets, and they introduced CREPE, a dataset with 25% of such questions, showing that current models struggle with factual correctness due to evidence retrieval challenges.
Information seeking users often pose questions with false presuppositions, especially when asking about unfamiliar topics. Most existing question answering (QA) datasets, in contrast, assume all questions have well defined answers. We introduce CREPE, a QA dataset containing a natural distribution of presupposition failures from online information-seeking forums. We find that 25% of questions contain false presuppositions, and provide annotations for these presuppositions and their corrections. Through extensive baseline experiments, we show that adaptations of existing open-domain QA models can find presuppositions moderately well, but struggle when predicting whether a presupposition is factually correct. This is in large part due to difficulty in retrieving relevant evidence passages from a large text corpus. CREPE provides a benchmark to study question answering in the wild, and our analyses provide avenues for future work in better modeling and further studying the task.