# Less is More: Mitigate Spurious Correlations for Open-Domain Dialogue Response Generation Models by Causal Discovery

arXiv: https://arxiv.org/abs/2303.01962 · HTML: https://mlsanity.com/paper/2303.01962/

## Key facts

- Cited by 10 papers in the map (data through 14 Oct 2026).
- Builds on 44 of its 65 references that are in the map.
- First posted on arXiv 2 Mar 2023.

## Builds on (most cited first, 30 of 44)

- [Adam: A Method for Stochastic Optimization](https://mlsanity.com/paper/1412.6980/) (2014)
- [PyTorch: An Imperative Style, High-Performance Deep Learning Library](https://mlsanity.com/paper/1912.01703/) (2019)
- [RoBERTa: A Robustly Optimized BERT Pretraining Approach](https://mlsanity.com/paper/1907.11692/) (2019)
- [Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks](https://mlsanity.com/paper/2005.11401/) (2020)
- [BERTScore: Evaluating Text Generation with BERT](https://mlsanity.com/paper/1904.09675/) (2019)
- [SQuAD: 100,000+ Questions for Machine Comprehension of Text](https://mlsanity.com/paper/1606.05250/) (2016)
- [Leveraging Passage Retrieval with Generative Models for Open Domain Question Answering](https://mlsanity.com/paper/2007.01282/) (2020)
- [A Diversity-Promoting Objective Function for Neural Conversation Models](https://mlsanity.com/paper/1510.03055/) (2015)
- [DialoGPT: Large-Scale Generative Pre-training for Conversational Response Generation](https://mlsanity.com/paper/1911.00536/) (2019)
- [Personalizing Dialogue Agents: I have a dog, do you have pets too?](https://mlsanity.com/paper/1801.07243/) (2018)
- [DailyDialog: A Manually Labelled Multi-turn Dialogue Dataset](https://mlsanity.com/paper/1710.03957/) (2017)
- [Recipes for building an open-domain chatbot](https://mlsanity.com/paper/2004.13637/) (2020)
- [Wizard of Wikipedia: Knowledge-Powered Conversational agents](https://mlsanity.com/paper/1811.01241/) (2018)
- [MAUVE: Measuring the Gap Between Neural Text and Human Text using Divergence Frontiers](https://mlsanity.com/paper/2102.01454/) (2021)
- [Confidence Regularized Self-Training](https://mlsanity.com/paper/1908.09822/) (2019)
- [RocketQA: An Optimized Training Approach to Dense Passage Retrieval for Open-Domain Question Answering](https://mlsanity.com/paper/2010.08191/) (2020)
- [How NOT To Evaluate Your Dialogue System: An Empirical Study of Unsupervised Evaluation Metrics for Dialogue Response Generation](https://mlsanity.com/paper/1603.08023/) (2016)
- [An Investigation of Why Overparameterization Exacerbates Spurious Correlations](https://mlsanity.com/paper/2005.04345/) (2020)
- [Towards Emotional Support Dialog Systems](https://mlsanity.com/paper/2106.01144/) (2021)
- [The Pushshift Reddit Dataset](https://mlsanity.com/paper/2001.08435/) (2020)
- [Revisiting Classifier Two-Sample Tests](https://mlsanity.com/paper/1610.06545/) (2016)
- [Beyond Goldfish Memory: Long-Term Open-Domain Conversation](https://mlsanity.com/paper/2107.07567/) (2021)
- [TransferTransfo: A Transfer Learning Approach for Neural Network Based Conversational Agents](https://mlsanity.com/paper/1901.08149/) (2019)
- [Emotional Chatting Machine: Emotional Conversation Generation with Internal and External Memory](https://mlsanity.com/paper/1704.01074/) (2017)
- [PLATO: Pre-trained Dialogue Generation Model with Discrete Latent Variable](https://mlsanity.com/paper/1910.07931/) (2019)
- [Unsupervised Domain Adaptation through Self-Supervision](https://mlsanity.com/paper/1909.11825/) (2019)
- [Best-Worst Scaling More Reliable than Rating Scales: A Case Study on Sentiment Intensity Annotation](https://mlsanity.com/paper/1712.01765/) (2017)
- [RankME: Reliable Human Ratings for Natural Language Generation](https://mlsanity.com/paper/1803.05928/) (2018)
- [Recognizing Emotion Cause in Conversations](https://mlsanity.com/paper/2012.11820/) (2020)
- [Do Neural Dialog Systems Use the Conversation History Effectively? An Empirical Study](https://mlsanity.com/paper/1906.01603/) (2019)

## Built on by (most cited first, 10 of 10)

- [RENOVI: A Benchmark Towards Remediating Norm Violations in Socio-Cultural Conversations](https://mlsanity.com/paper/2402.11178/) (2024)
- [LazyReview A Dataset for Uncovering Lazy Thinking in NLP Peer Reviews](https://mlsanity.com/paper/2504.11042/) (2025)
- [On the Reliability of Large Language Models for Causal Discovery](https://mlsanity.com/paper/2407.19638/) (2024)
- [Empathy Level Alignment via Reinforcement Learning for Empathetic Response Generation](https://mlsanity.com/paper/2408.02976/) (2024)
- [CausalScore: An Automatic Reference-Free Metric for Assessing Response Relevance in Open-Domain Dialogue Systems](https://mlsanity.com/paper/2406.17300/) (2024)
- [Learning in Order! A Sequential Strategy to Learn Invariant Features for Multimodal Sentiment Analysis](https://mlsanity.com/paper/2409.04473/) (2024)
- [IRIS: An Iterative and Integrated Framework for Verifiable Causal Discovery in the Absence of Tabular Data](https://mlsanity.com/paper/2510.09217/) (2025)
- [IMO: Greedy Layer-Wise Sparse Representation Learning for Out-of-Distribution Text Classification with Pre-trained Models](https://mlsanity.com/paper/2404.13504/) (2024)
- [Position Debiasing Fine-Tuning for Causal Perception in Long-Term Dialogue](https://mlsanity.com/paper/2406.02002/) (2024)
- [Reviewing the Reviewer: LLM-Assisted Reviewer Feedback Generation for Guideline Compliance](https://mlsanity.com/paper/2602.10118/) (2026)

## Abstract (arXiv)

In this paper, we conduct the first study on spurious correlations for open-domain response generation models based on a corpus CGDIALOG curated in our work. The cur rent models indeed suffer from spurious correlations and have a tendency of generating irrelevant and generic responses. Inspired by causal discovery algorithms, we propose a novel model-agnostic method for training and inference of response generation model using a conditional independence classifier. The classifier is trained by a constrained self-training method, coined CONSTRAIN, to overcome data scarcity. The experimental results based on both human and automatic evaluation show that our method significantly outperforms the competitive baselines in terms of relevance, informativeness, and fluency.