Research map / Large language models / Language, safety and interpretability
Teach Me to Explain: A Review of Datasets for Explainable Natural Language Processing
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Have annotators edit collected explanations rather than create them from scratch | NLI data collection · MNLI subset | annotation artifacts | Helped |
| Convert image captions into explanations without human editing or judging | VQA-E · Visual question answering | explanation plausibility | Hurt |
| Use non-comprehensive human highlights to evaluate comprehensive model highlights | Highlight explanation evaluation · Model-generated highlights | plausibility evaluation validity | Hurt |
| Collect comprehensive human highlights when evaluating comprehensive model highlights | Highlight explanation evaluation · Model-generated highlights | plausibility evaluation validity | Helped |
| Collect highlights for examples whose labels cannot be justified by highlighting | Highlight explanation datasets · Examples with labels not explainable by selected input spans | highlight sufficiency | Hurt |
| Embrace naturally occurring structure in explanations rather than discard template-like explanations | Free-text explanation collection · Tasks with naturally structured explanations | explanation informativeness | Helped |
| Use collect-and-edit quality control instead of relying only on initial collection | ExNLP dataset collection · Explanation data quality | data artifacts | Helped |
| Teach and test annotators on the underlying task and label set | ExNLP dataset collection · Annotation quality | data quality | Helped |
| Collect labels and explanations from the same annotators, or check annotator agreement with the assigned label | Post-hoc explanation collection · Explanation-label alignment | alignment with human decisions | Helped |
| Use a larger and more diverse pool of annotators | ExNLP dataset collection · Annotation representativeness | annotator bias | Helped |
| Collect multiple plausible explanations per instance | ExNLP datasets with subjective explanations · Explanation evaluation | evaluation fairness | Helped |
Closest papers by meaning
- e-SNLI: Natural Language Inference with Natural Language Explanations (NeurIPS 2018 · similarity 0.89)
- Towards Interpretable Natural Language Understanding with Explanations as Latent Variables (NeurIPS 2020 · similarity 0.88)
- Automated Assessment of Fidelity and Interpretability: An Evaluation Framework for Large Language Models’ Explanations (Student Abstract) (AAAI 2024 · similarity 0.86)
- Training Language Models to Explain Their Own Computations (NeurIPS 2026 · similarity 0.85)
- Faithful Explanations of Black-box NLP Models Using LLM-generated Counterfactuals (ICLR 2024 · similarity 0.84)
- Think in Latent, Explain in Language: Self-Explainable Latent Reasoning (ICML 2026 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.