Unsupervised Mutual Learning of Discourse Parsing and Topic Segmentation in Dialogue

Jiahui Xu, Feng Jiang, Anningzhe Gao, Luis Fernando D'Haro, Haizhou Li

Introduction

In the field of natural language processing, understanding the structural nuances of dialogue is essential for building sophisticated conversational agents. Two primary tasks facilitate this understanding: discourse parsing, which identifies the logical and rhetorical relationships between utterances (e.g., Question-Answer, Explanation), and topic segmentation, which identifies the boundaries where a conversation shifts from one subject to another.

Traditionally, these tasks have been treated as independent problems. Furthermore, the most effective models for these tasks typically rely on supervised learning, requiring vast amounts of human-annotated data that are both expensive and time-consuming to produce. This paper introduces the Unsupervised Mutual Learning Framework (UMLF), a novel approach that jointly addresses both tasks by leveraging their inherent linguistic interplay without the need for manual labels.

The Problem: Isolated and Data-Hungry Models

Existing dialogue systems face two significant hurdles. First, supervised models are constrained by the scarcity of labeled discourse and topic data. Second, current unsupervised methods often treat discourse and topic as isolated phenomena. This ignores a fundamental linguistic reality: rhetorical relations and topic structures are deeply intertwined. For example, a "Cause-Effect" rhetorical relation is highly likely to occur within a single topic, whereas a topic shift usually signals a break in rhetorical continuity. By ignoring this connection, models lose valuable signals that could improve structural accuracy.

Methodology: The UMLF Framework

The core innovation of this research is the Unsupervised Mutual Learning Framework (UMLF). The framework is built upon two linguistically grounded hypotheses:

  1. Local Discourse Coupling: Local rhetorical cues, such as a question followed by an immediate answer, reinforce the flow of a topic and help pinpoint where that flow is interrupted.
  2. Global Topology Constraint: The overall topic structure acts as a "skeleton" for the conversation, constraining where rhetorical relations are likely to exist. Relations are naturally denser within a topic and sparser across topic boundaries.

Architecture Overview

The UMLF architecture consists of three primary stages:

  • Representation Modules: The system utilizes Pre-trained Language Models (PLMs) like BART-Large alongside specialized encoders. Specifically, it employs SimCSE to ensure semantic consistency and NSP-BERT to evaluate conversational coherence, generating initial score matrices for both rhetoric and topics.
  • Mutual Learning Module: This module acts as the central engine of the framework. It uses local and global aggregators to fuse the rhetorical and topical matrices. Through a differentiable loss function, the module aligns these structures, ensuring they are semantically and structurally consistent with one another.
  • Parsing and Decoding: Finally, the framework applies classical decoding algorithms to the refined matrices. It uses TextTiling to determine topic boundaries and the Eisner algorithm to construct the final discourse trees.

Experimental Setup

To evaluate the robustness of UMLF, the researchers tested the framework across five diverse benchmarks:

  • Discourse Parsing: Evaluated on STAC (multi-party chats from an online game) and Molweni (a large-scale dialogue dataset based on Ubuntu technical support).
  • Topic Segmentation: Evaluated on Doc2Dial (goal-oriented dialogues), TIAGE (persona-based open-domain dialogue), and DialSeg711 (synthetic multi-topic dialogues).

Key Insights and Results

Mutual Reinforcement

Unlike previous "unidirectional" models where one task merely provides features for another, UMLF enables bidirectional reinforcement. Improving topic boundary detection helps the discourse parser filter out "noisy" or improbable links across topics. Conversely, identifying strong rhetorical clusters helps the topic segmenter recognize when a subject is still being discussed.

Performance Gains

UMLF demonstrated significant improvements over strong unsupervised baselines built on PLMs, showing a 12.32% to 14.33% increase in F1 scores for discourse parsing. Notably, despite having only 644 million parameters, the UMLF model achieved performance levels reaching approximately 90% of those seen in massive models like GPT-4o and Llama3-70B.

Scalability to LLMs

The framework is not limited to smaller encoders. When integrated with Large Language Models (LLMs) such as Llama3-8B and Qwen2-1.5B via LoRA fine-tuning, the mutual learning module significantly enhanced the models' ability to understand dialogue structure compared to standard text generation approaches.

Conclusion and Future Work

The Unsupervised Mutual Learning Framework proves that discourse and topic structures are more than the sum of their parts. By treating them as mutually reinforcing signals, UMLF achieves state-of-the-art unsupervised performance and remains robust even in long, complex dialogues where maintaining a global "topic skeleton" is critical.

Future research will focus on applying this framework to messier, real-world scenarios such as medical consultations or legal negotiations, where topics often overlap. Additionally, the researchers aim to explore how these improved structural representations can enhance downstream applications like automated dialogue summarization and more coherent response generation.

Summary updated

Abstract

In dialogue systems, discourse plays a crucial role in managing conversational focus and coordinating interactions. It consists of two key structures: rhetorical structure and topic structure. The former captures the logical flow of conversations, while the latter detects transitions between topics. Together, they improve the ability of a dialogue system to track conversation dynamics and generate contextually relevant high-quality responses. These structures are typically identified through discourse parsing and topic segmentation, respectively. However, existing supervised methods rely on costly manual annotations, while unsupervised methods often focus on a single task, overlooking the deep linguistic interplay between rhetorical and topic structures. To address these issues, we first introduce a unified representation that integrates rhetorical and topic structures, ensuring semantic consistency between them. Under the unified representation, we further propose two linguistically grounded hypotheses based on discourse theories: (1) Local Discourse Coupling, where rhetorical cues dynamically enhance topic-aware information flow, and (2) Global Topology Constraint, where topic structure patterns probabilistically constrain rhetorical relation distributions. Building on the unified representation and two hypotheses, we propose an unsupervised mutual learning framework (UMLF) that jointly models rhetorical and topic structures, allowing them to mutually reinforce each other without requiring additional annotations. We evaluate our approach on two rhetorical datasets (STAC and Molweni) and three topic segmentation datasets (DialSeg711, Doc2Dial, and TIAGE). Experimental results demonstrate that our method surpasses all strong baselines built on pre-trained language models (PLMs) and achieves competitive performance to state-of-the-art large language models (LLMs) such as GPT-4o. Furthermore, when applied to LLMs (Qwen2-1.5B-Instruct and Llama3-8B-Instruct), our framework achieves notable improvements, demonstrating its effectiveness in improving discourse structure modeling, even when integrated with powerful models.
BibTeX
@article{xu2025unsupervised,
  author = {Jiahui Xu and Feng Jiang and Anningzhe Gao and Luis Fernando D'Haro and Haizhou Li},
  title = {{Unsupervised Mutual Learning of Discourse Parsing and Topic Segmentation in Dialogue}},
  journal = {IEEE Transactions on Audio, Speech and Language Processing},
  year = {2025},
  url = {https://openreview.net/forum?id=onTJiM5SGh},
}
Figure