Assessing the Benchmarking Capacity of Machine Reading Comprehension Datasets

TL;DR

This paper introduces a semi-automated ablation-based method to evaluate whether MRC datasets truly test language understanding skills, revealing many questions can be answered without complex reasoning.

Contribution

The authors propose a novel methodology for assessing the benchmarking capacity of MRC datasets, highlighting the need for better dataset design to evaluate specific language understanding skills.

Findings

Most questions can be answered with content words only

Shuffled sentence words still allow high model performance

Current datasets may not effectively test complex reasoning skills

Abstract

Existing analysis work in machine reading comprehension (MRC) is largely concerned with evaluating the capabilities of systems. However, the capabilities of datasets are not assessed for benchmarking language understanding precisely. We propose a semi-automated, ablation-based methodology for this challenge; By checking whether questions can be solved even after removing features associated with a skill requisite for language understanding, we evaluate to what degree the questions do not require the skill. Experiments on 10 datasets (e.g., CoQA, SQuAD v2.0, and RACE) with a strong baseline model show that, for example, the relative scores of a baseline model provided with content words only and with shuffled sentence words in the context are on average 89.2% and 78.5% of the original score, respectively. These results suggest that most of the questions already answered correctly by the model do not necessarily require grammatical and complex reasoning. For precise benchmarking, MRC datasets will need to take extra care in their design to ensure that questions can correctly evaluate the intended skills.