Zero-Shot Cue-Grounded Topic Segmentation of Spoken Documents

2026-09-28

AI Summary

This study addresses the over-merging and over-segmentation issues in spoken document topic segmentation caused by granularity mismatches in existing large language models (LLMs). To this end, we propose a zero-shot cue-grounded segmentation framework. This approach introduces a novel unsupervised, training-free cue-grounding mechanism that leverages LLMs to prioritize explicit topic-initial cues for precise boundary localization, falling back to semantic segmentation when such cues are insufficient, thereby achieving multi-granularity adaptability. Experimental results demonstrate that the proposed method significantly outperforms baselines across six benchmarks. Furthermore, it exhibits strong robustness to noisy transcriptions and achieves high-precision topic segmentation with minimal API invocation costs.

Abstract

Topic segmentation structures spoken documents into coherent sections, facilitating navigation and downstream understanding. The appropriate granularity can vary substantially, ranging from broad thematic shifts to fine-grained subtopics. Existing LLM-based segmenters, however, often struggle to adapt to this variation, causing them to either merge distinct subtopics or over-segment coherent themes. To address this, we introduce Cue-Grounded Segmentation (CGS), a training-free framework that operates without any task-specific supervision. CGS first identifies phrases that explicitly signal the start of a new topic and uses their sentence positions as segment boundaries. When such cues are insufficient, it falls back to semantic segmentation, guided by the document structure inferred during cue extraction. Across six benchmarks and six LLM backbones, CGS consistently outperforms existing baselines, remains robust to noisy ASR transcripts, and achieves these gains with low API cost on proprietary models.

Problem

• Topic Segmentation
• Spoken Documents
• Granularity Adaptation
• Large Language Models

Innovation

• Zero-Shot Topic Segmentation
• Cue-Grounded Segmentation
• Training-Free Framework
• Spoken Documents
• Large Language Models