Saved in:
Bibliographic Details
Main Authors: Mackenzie, Pierre, Shah, Maya, Frenett, Patrick
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2601.03276
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915712460652544
author Mackenzie, Pierre
Shah, Maya
Frenett, Patrick
author_facet Mackenzie, Pierre
Shah, Maya
Frenett, Patrick
contents Topic segmentation using generative Large Language Models (LLMs) remains relatively unexplored. Previous methods use semantic similarity between sentences, but such models lack the long range dependencies and vast knowledge found in LLMs. In this work, we propose an overlapping and recursive prompting strategy using sentence enumeration. We also support the adoption of the boundary similarity evaluation metric. Results show that LLMs can be more effective segmenters than existing methods, but issues remain to be solved before they can be relied upon for topic segmentation.
format Preprint
id arxiv_https___arxiv_org_abs_2601_03276
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Topic Segmentation Using Generative Language Models
Mackenzie, Pierre
Shah, Maya
Frenett, Patrick
Computation and Language
Artificial Intelligence
Topic segmentation using generative Large Language Models (LLMs) remains relatively unexplored. Previous methods use semantic similarity between sentences, but such models lack the long range dependencies and vast knowledge found in LLMs. In this work, we propose an overlapping and recursive prompting strategy using sentence enumeration. We also support the adoption of the boundary similarity evaluation metric. Results show that LLMs can be more effective segmenters than existing methods, but issues remain to be solved before they can be relied upon for topic segmentation.
title Topic Segmentation Using Generative Language Models
topic Computation and Language
Artificial Intelligence
url https://arxiv.org/abs/2601.03276