Paper Reading

Reading lists for papers, including but not limited to Natural Language Processing (NLP).

• Deep Learning in NLP
• Debiasing
• Information Theory
• Dataset Construction
• Bias Suppression
• Others

---

Uncategorized

• Thinking Fair and Slow: "“Thinking” Fair and Slow: On the Efficacy of Structured Prompts for Debiasing Language Models". EMNLP(2024) [PDF]
• Regard Dataset: "The Woman Worked as a Babysitter: On Biases in Language Generation". EMNLP-IJCNLP(2019) [PDF][CODE]
• Regard V3: "Towards Controllable Biases in Language Generation". EMNLP-Findings(2020) [PDF][CODE]
• Bias in Bios: "Bias in Bios: A Case Study of Semantic Representation Bias in a High-Stakes Setting". FAT(2019) [PDF][CODE]
• Tree Structure Reasoning: "TREA: Tree-Structure Reasoning Schema for Conversational Recommendation". ACL(2023) [PDF][CODE]
• SEER: "SEER: Facilitating Structured Reasoning and Explanation via Reinforcement Learning". ACL(2024) [PDF][CODE]
• TrustLLM: "Trustllm: Trustworthiness in large language models". ICML(2024) [PDF][CODE][website]
• Entailment Trees: "Explaining Answers with Entailment Trees". EMNLP(2021) [PDF][CODE]
• LLM Data Annotation: "Large Language Models for Data Annotation and Synthesis: A Survey". EMNLP(2024) [PDF]
• Free-Text Explanation: "Reframing Human-AI Collaboration for Generating Free-Text Explanations". NAACL(2022) [PDF][CODE]
• Bias-NLI: "On Measuring and Mitigating Biased Inferences of Word Embeddings". AAAI(2019) [PDF]
• Gaps: "The Gaps between Pre-train and Downstream Settings in Bias Evaluation and Debiasing". COLING(2025) [PDF]
• Self-Correction: "The Capacity for Moral Self-Correction in Large Language Models". ArXiv(2023) [PDF]
• BBQ: "BBQ: A Hand-Built Bias Benchmark for Question Answering ". ACL-Findings(2022) [PDF][CODE]
• BNLI: "Evaluating gender bias of pre-trained language models in natural language inference by considering all labels". LREC-COLING(2024) [PDF]
• Adversarial NLI: "Adversarial NLI: A New Benchmark for Natural Language Understanding". ACL(2020) [PDF]
• SEAT: "On Measuring Social Biases in Sentence Encoders". NAACL(2019) [PDF]
• Implicit Rangking "A Study of Implicit Ranking Unfairness in Large Language Models". EMNLP Findings(2024) [PDF] [CODE]
• OCCUGENDER "Causally Testing Gender Bias in LLMs: A Case Study on Occupational Bias". NeurIPS(2024) [PDF] [CODE]
• Salmon Paper "Stereotyping Norwegian salmon: An inventory of pitfalls in fairness benchmark datasets". ACL(2021) [PDF]
• Romantic Relationship Prediction "On the Influence of Gender and Race in Romantic Relationship Prediction from Large Language Models". EMNLP(2024) [PDF]
• Theory-Grounded "Theory-Grounded Measurement of U.S. Social Stereotypes in English Language Models". NAACL(2022) [PDF]
• Marked Personas "Marked Personas: Using Natural Language Prompts to Measure Stereotypes in Language Models". ACL(2024) [PDF]
• SODAPOP "SODAPOP: Open-Ended Discovery of Social Biases in Social Commonsense Reasoning Models". EACL(2023) [PDF]
• DDRel Dataset "DDRel: A New Dataset for Interpersonal Relation Classification in Dyadic Dialogues" AAAI(2021) [PDF]
• Hiring Decisions "Do Large Language Models Discriminate in Hiring Decisions on the Basis of Race, Ethnicity, and Gender?". ACL(2024) [PDF]
• First Name Biases "Nichelle and Nancy: The Influence of Demographic Attributes and Tokenization Length on First Name Biases" ACL(2023) [PDF]
• ORPO "ORPO: Monolithic Preference Optimization without Reference Model". EMNLP(2024) [PDF]
• OPT "OPT: Open Pre-trained Transformer Language Models". ArXiv(2022) [PDF]
• Word Meanings Adapt "Understanding the Semantic Space: How Word Meanings Dynamically Adapt in the Context of a Sentence". SemSpace(2021) [PDF]
• Answer is all you need "Answer is All You Need: Instruction-following Text Embedding via Answering the Question" ACL(2024) [PDF]
• Social IQa "Social IQa: Commonsense Reasoning about Social Interactions". EMNLP-IJCNLP(2019) [PDF] [CODE]
• Symbolic Knowledge Distillation "Symbolic Knowledge Distillation: from General Language Models to Commonsense Models". NAACL(2022) [PDF]
• Implicit User Intention Understanding "Tell Me More! Towards Implicit User Intention Understanding of Language Model Driven Agents". ACL(2024) [PDF]
• SOCIAL IQA "SOCIAL IQA: Commonsense Reasoning about Social Interactions". EMNLP-IJCNLP(2019) [PDF]
• Low Frequency Names "Low Frequency Names Exhibit Bias and Overfitting in Contextualizing Language Models". EMNLP(2021) [PDF]
• Measure Fainess in Generative Models "On Measuring Fairness in Generative Models". NeurIPS(2023) [PDF]
• Dependency-Based Semantic Space "Dependency-Based Construction of Semantic Space Models". CL(2002) [PDF]
• Sentence BERT "Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks" ArXiv(2019) [PDF]
• Biased or Flawed "Biased or Flawed? Mitigating Stereotypes in Generative Language Models by Addressing Task-Specific Flaws" ArXiv(2024) [PDF]
• Bias Vector "Bias Vector: Mitigating Biases in Language Models with Task Arithmetic Approach" ArXiv(2024) [PDF]
• Multi-Objective Approach "Mitigating Social Bias in Large Language Models: A Multi-Objective Approach within a Multi-Agent Framework" ArXiv(2024) [PDF]
• Humans or LLMs as the Judge "Humans or LLMs as the Judge? A Study on Judgement Bias" EMNLP(2024) [PDF]
• Personas Is Not Helpful "When "A Helpful Assistant" Is Not Really Helpful: Personas in System Prompts Do Not Improve Performances of Large Language Models" [PDF]
• Measuring "Ask LLMs Directly, “What shapes your bias? ”: Measuring Social Bias in Large Language Models"
• MBTI Evaluation "Do LLMs Possess a Personality? Making the MBTI Test an Amazing Evaluation for Large Language Models"
• MBTI by LLMs "Can ChatGPT Assess Human Personalities? A General Evaluation Framework"
• MBTI Detection "Can Large Language Models Understand You Better? An MBTI Personality Detection Dataset Aligned with Population Traits"
• Role play "When “A Helpful Assistant” Is Not Really Helpful: Personas in System Prompts Do Not Improve Performances of Large Language Models"
• ProSA "ProSA: Assessing and Understanding the Prompt Sensitivity of LLMs"
• Quantifying Sensitivity "Quantifying Language Models' Sensitivity to Spurious Features in Prompt Design or: How I learned to start worrying about prompt formatting"
• In-Context Impersonation "In-Context Impersonation Reveals Large Language Models' Strengths and Biases"
• VISBIAS "VISBIAS: Measuring Explicit and Implicit Social Biases in Vision Language Models"
• Two Hundred Sentiment "Examining Gender and Race Bias in Two Hundred Sentiment Analysis Systems"
• No LLM is Free "No LLM is Free From Bias: A Comprehensive Study of Bias Evaluation in Large Language models"
• Gender Bias on Wikipedia "Men Are Elected, Women Are Married: Events Gender Bias on Wikipedia"
• Clinical Bias "CLIMB: A Benchmark of Clinical Bias in Large Language Models"
• Personal Biases "Are personalized stochastic parrots more dangerous? evaluating persona biases in dialogue systems"
• Effect Compression "Understanding the Effect of Model Compression on Social Bias in Large Language Models"
• Uncertainty Estimation "Uncertainty Estimation for Debiased Models: Does Fairness Hurt Reliability?"
• Mitigating Bias "Mitigating Word Bias in Zero-shot Prompt-based Classifiers"
• BiasAlert "BiasAlert: A Plug-and-play Tool for Social Bias Detection in LLMs"
• Short-Term "Evaluating Short-Term Temporal Fluctuations of Social Biases in Social Media Data and Masked Language Models"
• Cross-Lingual Training "Cross-Lingual Training for Automatic Question Generation"
• BiasDPO "BiasDPO: Mitigating Bias in Language Models through Direct Preference Optimization"
• ReSS "Your Stereotypical Mileage may Vary: Practical Challenges of Evaluating Biases in Multiple Languages and Cultural Contexts"
• Cross-Lingual Transfer "Cross-Lingual Transfer of Debiasing and Detoxification in Multilingual LLMs: An Extensive Investigation"
• Social Bias "Do LLMs Exhibit Human-like Response Biases? A Case Study in Survey Design"
• MBBQ "MBBQ: A Dataset for Cross-Lingual Comparison of Stereotypes in Generative LLMs"
• DPO and Toxicity "A Mechanistic Understanding of Alignment Algorithms: A Case Study on DPO and Toxicity"
• Across Languages "Preference Tuning For Toxicity Mitigation Generalizes Across Languages"

Deep Learning in NLP

• Data Augmentation: "A Survey of Data Augmentation Approaches for NLP". ACL-Findings(2021) [PDF]
• RAG: "Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks". NeurIPS(2020) [PDF]
• DPR: "Dense Passage Retrieval for Open-Domain Question Answering". EMNLP(2020) [PDF]

Debiasing

Survey

• Empirical Survey: "An Empirical Survey of the Effectiveness of Debiasing Techniques for Pre-trained Language Models". ACL(2022) [PDF]
• Bias and Fairness in LLMs "Bias and Fairness in Large Language Models: A Survey". CL(2024) [PDF]

Measure

• Measure Biases&Harms: "On Measures of Biases and Harms in NLP". AACL(2022) [PDF][CODE]

Static Embeddings

• Gender-Neutral: "Gender-preserving Debiasing for Pre-trained Word Embeddings". ACL(2019) [PDF] [CODE]

Pre-trained Language Models

• Attention-Debiasing: "Debiasing Pretrained Text Encoders by Paying Attention to Paying Attention". EMNLP(2022) [PDF] [CODE]
• Debiasing Masks: "Debiasing Masks: A New Framework for Shortcut Mitigation in NLU". EMNLP(2022) [PDF] [CODE]
• DebiasGAN: "DebiasGAN: Eliminating Position Bias in News Recommendation with Adversarial Learning". EMNLP(2022) [PDF] [CODE]
• DCLR: "Debiased Contrastive Learning of Unsupervised Sentence Representations". ACL(2022) [PDF] [CODE]
• Self-Debiasing: "Self-Diagnosis and Self-Debiasing: A Proposal for Reducing Corpus-Based Bias in NLP". TACL(2021) [PDF] [CODE]
• SENT-DEBIAS: "Towards Debiasing Sentence Representations". TACL(2021) [PDF] [CODE]
• FairFil: "FairFil: Contrastive Neural Debiasing Method for Pretrained Text Encoders". ICLR(2021) [PDF]
• Context-Debias: "Debiasing Pre-trained Contextualised Embeddings". EACL(2021) [PDF] [CODE]
• Occupation Data: "Good Secretaries, Bad Truck Drivers? Occupational Gender Stereotypes in Sentiment Analysis". EACL(2021) [PDF] [CODE]

Large Language Models

• Few-Shot Data Interventions: "Language Models Get a Gender Makeover: Mitigating Gender Bias with Few-Shot Data Interventions". ACL(2023) [PDF][CODE]
• Label Bias in LLMs: "Beyond Performance: Quantifying and Mitigating Label Bias in LLMs". NAACL(2024) [PDF][CODE]
• Conceptor-Aided Debiasing: "Conceptor-Aided Debiasing of Large Language Models". EMNLP(2023) [PDF]
• Likelihood-based Mitigation: "Likelihood-based Mitigation of Evaluation Bias in Large Language Models". ACL(2024) [PDF][CODE]

Information Theory

• Embedding-MI: "Estimating Mutual Information Between Dense Word Embeddings". ACL(2020) [PDF] [CODE]
• MI-Max: "A Mutual Information Maximization Perspective of Language Representation Learning". ACL(2020) [PDF] [CODE]

Dataset Construction

• Coscript: "Distilling Script Knowledge from Large Language Models for Constrained Language Planning". ACL(2023) [PDF] [CODE]
• Role of Demonstrations: "Rethinking the Role of Demonstrations: What Makes In-Context Learning Work?". EMNLP(2022) [PDF] [CODE]
• Order Sensitivity: "Fantastically Ordered Prompts and Where to Find Them: Overcoming Few-Shot Prompt Order Sensitivity". ACL(2022) [PDF]
• Holistic Descriptor Dataset: "I'm sorry to hear that: Finding New Biases in Language Models with a Holistic Descriptor Dataset". EMNLP(2022) [PDF] [CODE]

Bias Suppression

• Prompt Bias Suppression: "In-Contextual Gender Bias Suppression for Large Language Models". Findings of EACL(2023) [PDF] [CODE]

Impicit Bias

• Explicit and Implicit Gender: "Probing Explicit and Implicit Gender Bias through LLM Conditional Text Generation". Findings of EACL(2023) [PDF]

Others

• Wiki Bias: "Men Are Elected, Women Are Married: Events Gender Bias on Wikipedia". ACL(2021) [PDF] [CODE]
• Annotator Demographics Matter: "When Do Annotator Demographics Matter? Measuring the Influence of Annotator Demographics with the POPQUORN Dataset". LAW-XVII@ACL(2023) [PDF] [CODE]
• Label Debias: "Beyond Performance: Quantifying and Mitigating Label Bias in LLMs". NAACL(2024) [PDF] [CODE]

Back to Top