Improving Topic Modeling of Social Media Short Texts with Rephrasing: A Case Study of COVID-19 Related Tweets

2025-10-20

AI Summary

Short, informal, and noisy social media texts (e.g., tweets) severely degrade the coherence, redundancy, and interpretability of traditional topic models such as LDA. To address this, we propose TM-Rephrase—a model-agnostic, large language model (LLM)-driven rewriting framework that supports two strategies: generic rewriting and colloquial-to-formal transformation, both preserving semantic fidelity while normalizing input tweets. Experiments on 25,027 pandemic-related tweets demonstrate that preprocessing with TM-Rephrase significantly improves topic coherence, uniqueness, and diversity for downstream topic models (e.g., LDA), while reducing topic redundancy; the colloquial-to-formal strategy yields the strongest gains. This work pioneers the integration of style-controllable LLM rewriting into topic modeling pre-processing for short texts, establishing a scalable, plug-and-play paradigm for topic discovery in low-resource, high-noise domains.

Technology Category

• Natural Language Processing: (Large) Language Models
• Machine Learning: Large Multimodal Models (LMMs)
• Data Mining & Knowledge Management: Conversational Systems for Recommendation & Retrieval

Application Category

• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Search and Retrieval-Augmented AI: Large language models for search
• Web Mining and Content Analysis: Topic discovery and tracking

Abstract

Social media platforms such as Twitter (now X) provide rich data for analyzing public discourse, especially during crises such as the COVID-19 pandemic. However, the brevity, informality, and noise of social media short texts often hinder the effectiveness of traditional topic modeling, producing incoherent or redundant topics that are often difficult to interpret. To address these challenges, we have developed emph{TM-Rephrase}, a model-agnostic framework that leverages large language models (LLMs) to rephrase raw tweets into more standardized and formal language prior to topic modeling. Using a dataset of 25,027 COVID-19-related Twitter posts, we investigate the effects of two rephrasing strategies, general- and colloquial-to-formal-rephrasing, on multiple topic modeling methods. Results demonstrate that emph{TM-Rephrase} improves three metrics measuring topic modeling performance (i.e., topic coherence, topic uniqueness, and topic diversity) while reducing topic redundancy of most topic modeling algorithms, with the colloquial-to-formal strategy yielding the greatest performance gains and especially for the Latent Dirichlet Allocation (LDA) algorithm. This study contributes to a model-agnostic approach to enhancing topic modeling in public health related social media analysis, with broad implications for improved understanding of public discourse in health crisis as well as other important domains.

Problem

• Enhancing topic modeling for short social media texts
• Addressing brevity and noise in COVID-19 tweet analysis
• Improving topic coherence and reducing redundancy via rephrasing

Innovation

• Rephrasing tweets using large language models
• Converting colloquial language to formal text
• Improving topic coherence and reducing redundancy