LogiQA: A Challenge Dataset for Machine Reading Comprehension with Logical Reasoning

TL;DR

LogiQA is a new dataset designed to evaluate machine reading comprehension with a focus on logical reasoning, revealing that current models lag behind humans and serving as a benchmark for advancing logical AI.

Contribution

The paper introduces LogiQA, a comprehensive dataset for testing logical reasoning in machine reading comprehension, filling a gap in existing datasets.

Findings

Neural models perform significantly worse than humans on LogiQA.

LogiQA covers multiple types of deductive reasoning.

The dataset is publicly available for research use.

Abstract

Machine reading is a fundamental task for testing the capability of natural language understanding, which is closely related to human cognition in many aspects. With the rising of deep learning techniques, algorithmic models rival human performances on simple QA, and thus increasingly challenging machine reading datasets have been proposed. Though various challenges such as evidence integration and commonsense knowledge have been integrated, one of the fundamental capabilities in human reading, namely logical reasoning, is not fully investigated. We build a comprehensive dataset, named LogiQA, which is sourced from expert-written questions for testing human Logical reasoning. It consists of 8,678 QA instances, covering multiple types of deductive reasoning. Results show that state-of-the-art neural models perform by far worse than human ceiling. Our dataset can also serve as a benchmark for reinvestigating logical AI under the deep learning NLP setting. The dataset is freely available at https://github.com/lgw863/LogiQA-dataset