mlsanityData to 14 Oct 2026

Explaining with Contrastive Phrasal Highlighting: A Case Study in Assisting Humans to Detect Translation Differences

arXiv 2312.01582 ↗First posted 4 Dec 2023cs.CLcited by 8

Not on the map. Why?

Abstract (arXiv)

Explainable NLP techniques primarily explain by answering "Which tokens in the input are responsible for this prediction?''. We argue that for NLP models that make predictions by comparing two input texts, it is more useful to explain by answering "What differences between the two inputs explain this prediction?''. We introduce a technique to generate contrastive highlights that explain the predictions of a semantic divergence model via phrase-alignment-guided erasure. We show that the resulting highlights match human rationales of cross-lingual semantic differences better than popular post-hoc saliency techniques and that they successfully help people detect fine-grained meaning differences in human translations and critical machine translation errors.

Key facts. Cited by 8 papers in the map (data through 14 Oct 2026). Builds on 40 of its 69 references that are in the map. First posted on arXiv 4 Dec 2023.

Builds on

40 of 69 references.

References · 40

  1. 2018BERT: Pre-training of Deep Bidirectional Transformers for Language Understandingcited by 40,399
  2. 2017A Unified Approach to Interpreting Model Predictionscited by 6,555
  3. 2015A large annotated corpus for learning natural language inferencecited by 1,847
  4. 2017Towards A Rigorous Science of Interpretable Machine Learningcited by 1,416
  5. 2017Explanation in Artificial Intelligence: Insights from the Social Sciencescited by 1,206
  6. 2020BLEURT: Learning Robust Metrics for Text Generationcited by 1,203
  7. 2017SmoothGrad: removing noise by adding noisecited by 1,109
  8. 2020COMET: A Neural Framework for MT Evaluationcited by 1,033
  9. 2019Attention is not Explanationcited by 929
  10. 2019Attention is not not Explanationcited by 636
  11. 2021The FLORES-101 Evaluation Benchmark for Low-Resource and Multilingual Machine Translationcited by 560
  12. 2019ERASER: A Benchmark to Evaluate Rationalized NLP Modelscited by 530
28 more
  1. 2016RETAIN: An Interpretable Predictive Model for Healthcare using Reverse Time Attention Mechanismcited by 349
  2. 2021Experts, Errors, and Context: A Large-Scale Study of Human Evaluation for Machine Translationcited by 338
  3. 2020Questioning the AI: Informing Design Practices for Explainable AI User Experiencescited by 253
  4. 2016Understanding Neural Networks through Representation Erasurecited by 224
  5. 2022CometKiwi: IST-Unbabel 2022 Submission for the Quality Estimation Shared Taskcited by 207
  6. 2020Evaluating Explainable AI: Which Algorithmic Explanations Help Users Predict Model Behavior?cited by 190
  7. 2020SimAlign: High Quality Word Alignments without Parallel Training Data using Static and Contextualized Embeddingscited by 156
  8. 2018Pathologies of Neural Models Make Interpretations Difficultcited by 140
  9. 2020Proxy Tasks and Subjective Measures Can Be Misleading in Evaluating Explainable AI Systemscited by 139
  10. 2021Teach Me to Explain: A Review of Datasets for Explainable Natural Language Processingcited by 126
  11. 2020Explaining NLP Models via Minimal Contrastive Editing (MiCE)cited by 106
  12. 2022Interpreting Language Models with Contrastive Explanationscited by 92
  13. 2020Aligning Faithful Interpretations with their Social Attributioncited by 79
  14. 2018Beyond Word Importance: Contextual Decomposition to Extract Interactions from LSTMscited by 67
  15. 2021Beyond Expertise and Roles: A Framework to Characterize the Stakeholders of Interpretable Machine Learning and their Needscited by 64
  16. 2016"What is Relevant in a Text Document?": An Interpretable Machine Learning Approachcited by 53
  17. 2021Prompting Contrastive Explanations for Commonsense Reasoning Taskscited by 45
  18. 2020Interpretation of NLP models through input marginalizationcited by 44
  19. 2020Gradient-based Analysis of NLP Models is Manipulablecited by 41
  20. 2020Linguistically-Informed Transformations (LIT): A Method for Automatically Generating Contrast Setscited by 31
  21. 2020Staying True to Your Word: (How) Can Attention Become Explanation?cited by 18
  22. 2020Detecting Fine-Grained Cross-Lingual Semantic Divergences without Supervision by Learning to Rankcited by 18
  23. 2021Neural semi-Markov CRF for Monolingual Word Alignmentcited by 16
  24. 2018Identifying Semantic Divergences in Parallel Text without Annotationscited by 9
  25. 2021Let the CAT out of the bag: Contrastive Attributed explanations for Textcited by 9
  26. 2019Interrogating the Explanatory Power of Attention in Neural Machine Translationcited by 7
  27. 2019Towards Recognizing Phrase Translation Processes: Experiments on English-Frenchcited by 3
  28. 2022Spanish Abstract Meaning Representation: Annotation of a General Corpuscited by 2

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2024: 1; 2025: 4; 2026: 3.

8 papers.

  1. 2025Translation in the Hands of Many:Centering Lay Users in Machine Translation Interactionscited by 8
  2. 2025An Interdisciplinary Approach to Human-Centered Machine Translationcited by 6
  3. 2025Should I Share this Translation? Evaluating Quality Feedback for User Reliance on Machine Translationcited by 5
  4. 2025A Necessary Step toward Faithfulness: Measuring and Improving Consistency in Free-Text Explanationscited by 3
  5. 2024A Context-aware Framework for Translation-mediated Conversationscited by 2
  6. 2026Contrastive ESA: Human Evaluation of Multiple Translations at Oncecited by 1
  7. 2026ChartAttack: Testing the Vulnerability of LLMs to Malicious Prompting in Chart Generationcited by 0
  8. 2026Measuring User's Mental Models of Speech Translation in Human-AI Collaborationcited by 0