Modeling Annotators: A Generative Approach to Learning from Annotator Rationales

A human annotator can provide hints to a machine learner by highlighting contextual "rationales" for each of his or her annotations (Zaidan et al., 2007). How can one exploit this side information to better learn the desired parameters θ? We present a generative model of how a given annotator, knowing the true θ, stochastically chooses rationales. Thus, observing the rationales helps us infer the true θ. We collect substring rationales for a sentiment classification task (Pang and Lee, 2004) and use them to obtain significant accuracy improvements for each annotator. Our new generative approach exploits the rationales more effectively than our previous "masking SVM" approach. It is also more principled, and could be adapted to help learn other kinds of probabilistic classifiers for quite different tasks.

• Publication year

  2008

• Venue

  Conference on Empirical Methods in Natural Language Processing

• Publication date

  2008-10-25

• Fields of study

  Computer Science

• Identifiers
  DOI 10.3115/1613715.1613721
• External record

  Open on Semantic Scholar

• Source metadata

  Semantic Scholar

• No claims are published for this paper.

• No concepts are published for this paper.

• Learning from labeled features using generalized expectation criteria

  2008cited by this paper
• An interactive algorithm for asking and incorporating feature feedback into support vector machines

  2007cited by this paper
• Using “Annotator Rationales” to Improve Machine Learning for Text Categorization

  2007influential reference
• Prototype-Driven Learning for Sequence Models

  2006cited by this paper
• Peekaboom: a game for locating objects in images

  2006cited by this paper
• Active Learning with Feedback on Features and Instances

  2006cited by this paper
• A Sentimental Education: Sentiment Analysis Using Subjectivity Summarization Based on Minimum Cuts

  2004cited by this paper
• Head-Driven Statistical Models for Natural Language Parsing

  2003cited by this paper
• Thumbs up? Sentiment Classification using Machine Learning Techniques

  2002cited by this paper
• Conditional Random Fields: Probabilistic Models for Segmenting and Labeling Sequence Data

  2001cited by this paper
• Rule Writing or Annotation: Cost-efficient Resource Usage for Base Noun Phrase Chunking

  2000cited by this paper
• Man vs. Machine: A Case Study in Base Noun Phrase Learning

  1999cited by this paper

• X-MuTeST: A Multilingual Benchmark for Explainable Hate Speech Detection and A Novel LLM-consulted Explanation Framework

  2026cites this paper
• Explainable Document Classification via Concept Whitening and Stable Graph Patterns

  2025cites this paper
• MFTCXplain: A Multilingual Benchmark Dataset for Evaluating the Moral Reasoning of LLMs through Multi-hop Hate Speech Explanation

  2025cites this paper
• Explainable Sentiment Analysis through Counterfactual Reasoning

  2025cites this paper
• EvalxNLP: A Framework for Benchmarking Post-Hoc Explainability Methods on NLP Models

  2025cites this paper
• AutoTag: a framework for generating training data

  2025cites this paper
• MFTCXplain: A Multilingual Benchmark Dataset for Evaluating the Moral Reasoning of LLMs through Hate Speech Multi-hop Explanation

  2025cites this paper
• Policy-based Word Subset Selection for Explaining Black-Box Language Models

  2025cites this paper
• Latent Concept-based Explanation of NLP Models

  2024cites this paper
• A simple and efficient filter feature selection method via document-term matrix unitization

  2024cites this paper
• TriSum: Learning Summarization Ability from Large Language Models with Structured Rationale

  2024cites this paper
• Beyond explaining: XAI-based Adaptive Learning with SHAP Clustering for Energy Consumption Prediction

  2024cites this paper
• GradToken: Decoupling tokens with class-aware gradient for visual explanation of Transformer network

  2024cites this paper
• How to Train Text Summarization Model with Weak Supervisions

  2024cites this paper
• Explanation Regularisation through the Lens of Attributions

  2024influential citation
• Evaluating the Reliability of Self-Explanations in Large Language Models

  2024cites this paper
• Fast Explanations via Policy Gradient-Optimized Explainer

  2024cites this paper
• Human-annotated rationales and explainable text classification: a survey

  2024cites this paper
• A Marketplace Price Anomaly Detection System at Scale

  2023cites this paper
• M4: A Unified XAI Benchmark for Faithfulness Evaluation of Feature Attribution Methods across Metrics, Modalities and Models

  2023cites this paper
• Hybrid Crowd-AI Learning for Human-Interpretable Symbolic Rules in Image Classification

  2023cites this paper
• Explaining Data Patterns in Natural Language with Language Models

  2023cites this paper
• Semantic HELM: A Human-Readable Memory for Reinforcement Learning

  2023cites this paper
• Make Your Decision Convincing! A Unified Two-Stage Framework: Self-Attribution and Decision-Making

  2023cites this paper
• SMOOT: Saliency Guided Mask Optimized Online Training

  2023cites this paper
• Goodhart’s Law Applies to NLP’s Explanation Benchmarks

  2023cites this paper
• LEXPLAIN: Improving Model Explanations via Lexicon Supervision

  2023cites this paper
• A Dataset for Explainable Sentiment Analysis in the German Automotive Industry

  2023influential citation
• Recipe-MPR: A Test Collection for Evaluating Multi-aspect Preference-based Natural Language Retrieval

  2023cites this paper
• Logistic Regression Matching Pursuit algorithm for text classification

  2023cites this paper
• HuCurl: Human-induced Curriculum Discovery

  2023cites this paper
• Semantic HELM: An Interpretable Memory for Reinforcement Learning

  2023cites this paper
• Unsupervised Selective Rationalization with Noise Injection

  2023cites this paper
• Explaining black box text modules in natural language with language models

  2023cites this paper
• ClaimDiff: Comparing and Contrasting Claims on Contentious Issues

  2022cites this paper
• Beyond Explaining: Opportunities and Challenges of XAI-Based Model Improvement

  2022cites this paper
• Natural Language Descriptions of Deep Visual Features

  2022cites this paper
• DARE: Disentanglement-Augmented Rationale Extraction

  2022cites this paper
• Augmenting interpretable models with large language models during training

  2022cites this paper
• StyLEx: Explaining Style Using Human Lexical Annotations

  2022cites this paper
• ER-Test: Evaluating Explanation Regularization Methods for Language Models

  2022influential citation
• Are Multilingual Sentiment Models Equally Right for the Right Reasons?

  2022cites this paper
• Explaining Patterns in Data with Language Models via Interpretable Autoprompting

  2022cites this paper
• Explainable Text Classification Techniques in Legal Document Review: Locating Rationales without Using Human Annotated Training Text Snippets

  2022cites this paper
• KNIFE: Knowledge Distillation with Free-Text Rationales

  2022cites this paper
• Going Beyond XAI: A Systematic Survey for Explanation-Guided Learning

  2022influential citation
• The First Workshop on Learning with Natural Language Supervision

  2022cites this paper
• XMD: An End-to-End Framework for Interactive Explanation-Based Debugging of NLP Models

  2022cites this paper
• StyLEx: Explaining Styles with Lexicon-Based Human Perception

  2022cites this paper
• FLUTE: Figurative Language Understanding through Textual Explanations

  2022cites this paper
• Adaptive Ranking-based Sample Selection for Weakly Supervised Class-imbalanced Text Classification

  2022cites this paper
• Explaining Patterns in Data with Language Models via Interpretable Autoprompting

  2022cites this paper
• ferret: a Framework for Benchmarking Explainers on Transformers

  2022cites this paper
• Materials information extraction via automatically generated corpus

  2022cites this paper
• Explaining Why: How Instructions and User Interfaces Impact Annotator Rationales When Labeling Text Data

  2022cites this paper
• Multiple weak supervision for short text classification

  2022cites this paper
• BAGEL: A Benchmark for Assessing Graph Neural Network Explanations

  2022cites this paper
• Fixing Model Bugs with Natural Language Patches

  2022cites this paper
• Fixing Model Bugs with Natural Language Patches

  2022cites this paper
• ER-TEST Evaluating Explanation Regularization Methods for NLP Models

  2022influential citation
• A Fine-grained Interpretability Evaluation Benchmark for Neural NLP

  2022cites this paper
• Ranking-Constrained Learning with Rationales for Text Classification

  2022cites this paper
• A survey on improving NLP models with human explanations

  2022cites this paper
• AdapLeR: Speeding up Inference by Adaptive Length Reduction

  2022cites this paper
• Are Shortest Rationales the Best Explanations for Human Understanding?

  2022cites this paper
• Knowledge-Grounded Self-Rationalization via Extractive and Natural Language Explanations

  2021cites this paper
• Explain and Predict, and then Predict Again

  2021influential citation
• Knowledge-Guided Sentiment Analysis Via Learning From Natural Language Explanations

  2021cites this paper
• SELFEXPLAIN: A Self-Explaining Architecture for Neural Text Classifiers

  2021cites this paper
• On the Faithfulness Measurements for Model Interpretations

  2021cites this paper
• Rationale-Inspired Natural Language Explanations with Commonsense

  2021cites this paper
• Rational LAMOL: A Rationale-based Lifelong Learning Framework

  2021cites this paper
• AutoTriggER: Named Entity Recognition with Auxiliary Trigger Extraction

  2021cites this paper
• Does External Knowledge Help Explainable Natural Language Inference? Automatic Evaluation vs. Human Ratings

  2021cites this paper
• WRENCH: A Comprehensive Benchmark for Weak Supervision

  2021cites this paper
• Improving Deep Learning Interpretability by Saliency Guided Training

  2021cites this paper
• UNIREX: A Unified Learning Framework for Language Model Rationale Extraction

  2021influential citation
• Unifying Model Explainability and Robustness for Joint Text Classification and Rationale Extraction

  2021cites this paper
• On the Sensitivity and Stability of Model Interpretations in NLP

  2021cites this paper
• AutoTriggER: Label-Efficient and Robust Named Entity Recognition with Auxiliary Trigger Extraction

  2021cites this paper
• Transformer Interpretability Beyond Attention Visualization

  2020cites this paper
• Robustness to Spurious Correlations in Text Classification via Automatically Generated Counterfactuals

  2020cites this paper
• FiD-Ex: Improving Sequence-to-Sequence Models for Extractive Rationale Generation

  2020cites this paper
• Machine-annotated Rationales: Faithfully Explaining Text Classification

  2020cites this paper
• ExpBERT: Representation Engineering with Natural Language Explanations

  2020cites this paper
• Learning to Faithfully Rationalize by Construction

  2020cites this paper
• WT5?! Training Text-to-Text Models to Explain their Predictions

  2020influential citation
• Towards Prediction Explainability through Sparse Communication

  2020cites this paper
• Towards Faithfully Interpretable NLP Systems: How Should We Define and Evaluate Faithfulness?

  2020cites this paper
• Fast and Three-rious: Speeding Up Weak Supervision with Triplet Methods

  2020cites this paper
• Consumer-Driven Explanations for Machine Learning Decisions: An Empirical Study of Robustness

  2020cites this paper
• Active neural learners for text with dual supervision

  2020cites this paper
• Aligning Faithful Interpretations with their Social Attribution

  2020cites this paper
• Concept Matching for Low-Resource Classification

  2020cites this paper
• Cost-effective Interactive Attention Learning with Neural Attention Processes

  2020cites this paper
• Train and You'll Miss It: Interactive Model Iteration with Weak Supervision and Pre-Trained Embeddings

  2020cites this paper
• Data Programming by Demonstration: A Framework for Interactively Learning Labeling Functions

  2020cites this paper
• Annotator Rationales for Labeling Tasks in Crowdsourcing

  2020cites this paper
• ALICE: Active Learning with Contrastive Natural Language Explanations

  2020cites this paper
• Explaining The Efficacy of Counterfactually-Augmented Data

  2020cites this paper