mlsanityData to 13 Oct 2026

Large-scale Cloze Test Dataset Created by Teachers

arXiv 1711.03225 ↗First posted 9 Nov 2017cs.CL cs.AIcited by 27

Not on the map. Why?

Abstract (arXiv)

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.

Key facts. Cited by 27 papers in the map (data through 13 Oct 2026). First posted on arXiv 9 Nov 2017.

Builds on

No reference list yet.

Built on by

Cited by year (2026 to 13 Oct)

Citing papers per year: 2021: 5; 2022: 5; 2023: 4; 2024: 7; 2025: 3; 2026: 3.

27 papers.

Sort:

By area:

  1. 2024Benchmark Data Contamination of Large Language Models: A Surveycited by 145
  2. 2022Fantastic Questions and Where to Find Them: FairytaleQA -- An Authentic Dataset for Narrative Comprehensioncited by 80
  3. 2024Protecting Privacy in Multimodal Large Language Models with MLLMU-Benchcited by 62
  4. 2021Using Adversarial Attacks to Reveal the Statistical Bias in Machine Reading Comprehension Modelscited by 29
  5. 2023On the application of Large Language Models for language teaching and assessment technologycited by 27
  6. 2021English Machine Reading Comprehension Datasets: A Surveycited by 26
  7. 2024Distractor Generation in Multiple-Choice Tasks: A Survey of Methods, Datasets, and Evaluationcited by 25
  8. 2025Modality-Aware Neuron Pruning for Unlearning in Multimodal Large Language Modelscited by 25
  9. 2022EduQG: A Multi-format Multiple Choice Dataset for the Educational Domaincited by 11
  10. 2022Pretraining without Wordpieces: Learning Over a Vocabulary of Millions of Wordscited by 6
  11. 2021A Survey on non-English Question Answering Datasetcited by 5
  12. 2022Constructing Open Cloze Tests Using Generation and Discrimination Capabilities of Transformerscited by 5
15 more
  1. 2023ViMMRC 2.0 -- Enhancing Machine Reading Comprehension on Vietnamese Literature Textcited by 5
  2. 2024CDGP: Automatic Cloze Distractor Generation based on Pre-trained Language Modelcited by 3
  3. 2021Native Chinese Reader: A Dataset Towards Native-Level Chinese Machine Reading Comprehensioncited by 2
  4. 2021LRG at SemEval-2021 Task 4: Improving Reading Comprehension with Abstract Words using Augmentation, Linguistic Features and Votingcited by 1
  5. 2023Explicit Alignment and Many-to-many Entailment Based Reasoning for Conversational Machine Readingcited by 1
  6. 2024Controlling Cloze-test Question Item Difficulty with PLM-based Surrogate Models for IRT Assessmentcited by 1
  7. 2024FairytaleQA Translated: Enabling Educational Question and Answer Generation in Less-Resourced Languagescited by 1
  8. 2022InDEX: Indonesian Idiom and Expression Dataset for Cloze Testcited by 0
  9. 2023Cross-corpus Readability Compatibility Assessment for English Textscited by 0
  10. 2024Recent Advances in Multi-Choice Machine Reading Comprehension: A Survey on Methods and Datasetscited by 0
  11. 2025Difficulty-Controllable Cloze Question Distractor Generationcited by 0
  12. 2025ESACT: An End-to-End Sparse Accelerator for Compute-Intensive Transformers via Local Similaritycited by 0
  13. 2026Gated Tree Cross-Attention for Checkpoint-Compatible Syntax Injection in Decoder-Only LLMscited by 0
  14. 2026Beetle: A Bilingual Model Suite for Modelling Second-Language Processingcited by 0
  15. 2026The Architect, the Adversary, and the Judge: Closed-Loop Generation of Standards-Aligned Assessment Items at Scalecited by 0