# Unsupervised Conversation Disentanglement through Co-Training

arXiv: https://arxiv.org/abs/2109.03199 · HTML: https://mlsanity.com/paper/2109.03199/

## Key facts

- Cited by 8 papers in the map (data through 14 Oct 2026).
- Builds on 11 of its 36 references that are in the map.
- First posted on arXiv 7 Sep 2021.

## Builds on (most cited first, 11 of 11)

- [BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding](https://mlsanity.com/paper/1810.04805/) (2018)
- [Adam: A Method for Stochastic Optimization](https://mlsanity.com/paper/1412.6980/) (2014)
- [DialoGPT: Large-Scale Generative Pre-training for Conversational Response Generation](https://mlsanity.com/paper/1911.00536/) (2019)
- [Co-training Embeddings of Knowledge Graphs and Entity Descriptions for Cross-lingual Entity Alignment](https://mlsanity.com/paper/1806.06478/) (2018)
- [Response Selection for Multi-Party Conversations with Dynamic Topic Tracking](https://mlsanity.com/paper/2010.07785/) (2020)
- [Multi-turn Response Selection using Dialogue Dependency Relations](https://mlsanity.com/paper/2010.01502/) (2020)
- [Structured Attention for Unsupervised Dialogue Structure Induction](https://mlsanity.com/paper/2009.08552/) (2020)
- [Unsupervised Dialog Structure Learning](https://mlsanity.com/paper/1904.03736/) (2019)
- [Who did They Respond to? Conversation Structure Modeling using Masked Hierarchical Transformer](https://mlsanity.com/paper/1911.10666/) (2019)
- [Online Conversation Disentanglement with Pointer Networks](https://mlsanity.com/paper/2010.11080/) (2020)
- [Reinforced Co-Training](https://mlsanity.com/paper/1804.06035/) (2018)

## Built on by (most cited first, 8 of 8)

- [Enhanced Speaker-aware Multi-party Multi-turn Dialogue Comprehension](https://mlsanity.com/paper/2109.04066/) (2021)
- [GIFT: Graph-Induced Fine-Tuning for Multi-Party Conversation Understanding](https://mlsanity.com/paper/2305.09360/) (2023)
- [Structural Characterization for Dialogue Disentanglement](https://mlsanity.com/paper/2110.08018/) (2021)
- [Multi-Party Conversational Agents: A Survey](https://mlsanity.com/paper/2505.18845/) (2025)
- [Conversation Disentanglement with Bi-Level Contrastive Learning](https://mlsanity.com/paper/2210.15265/) (2022)
- [CluCDD:Contrastive Dialogue Disentanglement via Clustering](https://mlsanity.com/paper/2302.08146/) (2023)
- [Multi-party Response Generation with Relation Disentanglement](https://mlsanity.com/paper/2403.10827/) (2024)
- [CO3: Low-resource Contrastive Co-training for Generative Conversational Query Rewrite](https://mlsanity.com/paper/2403.11873/) (2024)

## Abstract (arXiv)

Conversation disentanglement aims to separate intermingled messages into detached sessions, which is a fundamental task in understanding multi-party conversations. Existing work on conversation disentanglement relies heavily upon human-annotated datasets, which are expensive to obtain in practice. In this work, we explore to train a conversation disentanglement model without referencing any human annotations. Our method is built upon a deep co-training algorithm, which consists of two neural networks: a message-pair classifier and a session classifier. The former is responsible for retrieving local relations between two messages while the latter categorizes a message to a session by capturing context-aware information. Both networks are initialized respectively with pseudo data built from an unannotated corpus. During the deep co-training process, we use the session classifier as a reinforcement learning component to learn a session assigning policy by maximizing the local rewards given by the message-pair classifier. For the message-pair classifier, we enrich its training data by retrieving message pairs with high confidence from the disentangled sessions predicted by the session classifier. Experimental results on the large Movie Dialogue Dataset demonstrate that our proposed approach achieves competitive performance compared to the previous supervised methods. Further experiments show that the predicted disentangled conversations can promote the performance on the downstream task of multi-party response selection.