Break, Perturb, Build: Automatic Perturbation of Reasoning Paths through Question Decomposition

by   Mor Geva, et al.

Recent efforts to create challenge benchmarks that test the abilities of natural language understanding models have largely depended on human annotations. In this work, we introduce the "Break, Perturb, Build" (BPB) framework for automatic reasoning-oriented perturbation of question-answer pairs. BPB represents a question by decomposing it into the reasoning steps that are required to answer it, symbolically perturbs the decomposition, and then generates new question-answer pairs. We demonstrate the effectiveness of BPB by creating evaluation sets for three reading comprehension (RC) benchmarks, generating thousands of high-quality examples without human intervention. We evaluate a range of RC models on our evaluation sets, which reveals large performance gaps on generated examples compared to the original data. Moreover, symbolic perturbations enable fine-grained analysis of the strengths and limitations of models. Last, augmenting the training data with examples generated by BPB helps close performance gaps, without any drop on the original data distribution.



There are no comments yet.


page 2


VisualMRC: Machine Reading Comprehension on Document Images

Recent studies on machine reading comprehension have focused on text-lev...

Discrete Reasoning Templates for Natural Language Understanding

Reasoning about information from multiple parts of a passage to derive a...

Evaluating NLP Models via Contrast Sets

Standard test sets for supervised learning evaluate in-distribution gene...

ASQ: Automatically Generating Question-Answer Pairs using AMRs

In this work, we introduce ASQ, a tool to automatically mine questions a...

Natural Perturbation for Robust Question Answering

While recent models have achieved human-level scores on many NLP dataset...

Automatic Generation of Contrast Sets from Scene Graphs: Probing the Compositional Consistency of GQA

Recent works have shown that supervised models often exploit data artifa...
This week in AI

Get the week's most popular data science and artificial intelligence research sent straight to your inbox every Saturday.