FeTaQA: Free-form Table Question Answering
This dataset addresses the problem of evaluating complex reasoning in table question answering for AI researchers, though it is incremental as it builds on existing table QA work.
The authors introduced FeTaQA, a 10K-pair dataset for table question answering that requires generating free-form text answers from multiple discontinuous table facts, addressing limitations of existing datasets that focus on factual questions with short answers. They demonstrated the dataset's challenge by showing benchmark methods based on semantic parsing and pretrained text generation models both struggle with it.
Existing table question answering datasets contain abundant factual questions that primarily evaluate the query and schema comprehension capability of a system, but they fail to include questions that require complex reasoning and integration of information due to the constraint of the associated short-form answers. To address these issues and to demonstrate the full challenge of table question answering, we introduce FeTaQA, a new dataset with 10K Wikipedia-based {table, question, free-form answer, supporting table cells} pairs. FeTaQA yields a more challenging table question answering setting because it requires generating free-form text answers after retrieval, inference, and integration of multiple discontinuous facts from a structured knowledge source. Unlike datasets of generative QA over text in which answers are prevalent with copies of short text spans from the source, answers in our dataset are human-generated explanations involving entities and their high-level relations. We provide two benchmark methods for the proposed task: a pipeline method based on semantic-parsing-based QA systems and an end-to-end method based on large pretrained text generation models, and show that FeTaQA poses a challenge for both methods.