Large-scale Cloze Test Dataset Created by Teachers

Abstract

Cloze tests are widely adopted in language exams to evaluate students' language proficiency. In this paper, we propose the first large-scale human-created cloze test dataset CLOTH, containing questions used in middle-school and high-school language exams. With missing blanks carefully created by teachers and candidate choices purposely designed to be nuanced, CLOTH requires a deeper language understanding and a wider attention span than previously automatically-generated cloze datasets. We test the performance of dedicatedly designed baseline models including a language model trained on the One Billion Word Corpus and show humans outperform them by a significant margin. We investigate the source of the performance gap, trace model deficiencies to some distinct properties of CLOTH, and identify the limited ability of comprehending the long-term context to be the key bottleneck.

EMNLP 2018

References in corpus (6)

• Think you have Solved Question Answering? Try ARC, the AI2 Reasoning Challenge
• Aligning Books and Movies: Towards Story-like Visual Explanations by Watching Movies and Reading Books
• NewsQA: A Machine Comprehension Dataset
• Who did What: A Large-Scale Person-Centered Cloze Dataset
• Learning to Paraphrase for Question Answering
• Dynamic Fusion Networks for Machine Reading Comprehension

Cited by in corpus (3)

• Machine Reading Comprehension: a Literature Review
• Efficient and Robust Question Answering from Minimal Context over Documents
• Pre-training Text Representations as Meta Learning