FREB-TQA: A Fine-Grained Robustness Evaluation Benchmark for Table Question Answering
This addresses the need for fine-grained robustness evaluation in TQA, which is crucial for developing reliable systems, though it is incremental as it builds on prior work on TQA robustness.
The paper tackles the problem of evaluating robustness in Table Question Answering (TQA) systems by formalizing three desiderata and creating a novel benchmark, revealing that no state-of-the-art TQA system consistently excels across these aspects.
Table Question Answering (TQA) aims at composing an answer to a question based on tabular data. While prior research has shown that TQA models lack robustness, understanding the underlying cause and nature of this issue remains predominantly unclear, posing a significant obstacle to the development of robust TQA systems. In this paper, we formalize three major desiderata for a fine-grained evaluation of robustness of TQA systems. They should (i) answer questions regardless of alterations in table structure, (ii) base their responses on the content of relevant cells rather than on biases, and (iii) demonstrate robust numerical reasoning capabilities. To investigate these aspects, we create and publish a novel TQA evaluation benchmark in English. Our extensive experimental analysis reveals that none of the examined state-of-the-art TQA systems consistently excels in these three aspects. Our benchmark is a crucial instrument for monitoring the behavior of TQA systems and paves the way for the development of robust TQA systems. We release our benchmark publicly.