WIQA: A dataset for "What if..." reasoning over procedural text

Abstract

We introduce WIQA, the first large-scale dataset of "What if..." questions over procedural text. WIQA contains three parts: a collection of paragraphs each describing a process, e.g., beach erosion; a set of crowdsourced influence graphs for each paragraph, describing how one change affects another; and a large (40k) collection of "What if...?" multiple-choice questions derived from the graphs. For example, given a paragraph about beach erosion, would stormy weather result in more or less erosion (or have no effect)? The task is to answer the questions, given their associated paragraph. WIQA contains three kinds of questions: perturbations to steps mentioned in the paragraph; external (out-of-paragraph) perturbations requiring commonsense knowledge; and irrelevant (no effect) perturbations. We find that state-of-the-art models achieve 73.8% accuracy, well below the human performance of 96.3%. We analyze the challenges, in particular tracking chains of influences, and present the dataset as an open challenge to the community.

arXiv commentAccepted at EMNLP 2019

References in corpus (2)

• Tracking the World State with Recurrent Entity Networks
• Reasoning about Actions and State Changes by Injecting Commonsense Knowledge

Cited by in corpus (4)

• Back to the Future: Unsupervised Backprop-based Decoding for Counterfactual and Abductive Commonsense Reasoning
• Beyond Leaderboards: A survey of methods for revealing weaknesses in Natural Language Inference data and models
• EIGEN: Event Influence GENeration using Pre-trained Language Models
• Logic-Guided Data Augmentation and Regularization for Consistent Question Answering