TWIST: Training-free and Label-free Short Text Clustering through Iterative Vector Updating with LLMs

2025-10-08

AI Summary

Intent clustering of short texts in customer service scenarios faces two key challenges: the absence of labeled data and unknown numbers of clusters. To address these, we propose a lightweight, training-free, label-free clustering method that leverages large language models (LLMs) for semantic guidance. Specifically, it iteratively refines initial embeddings via sparse vector updates—enhancing embedding quality without requiring domain-specific prior knowledge. The approach is model-agnostic, compatible with arbitrary embedding models and downstream clustering algorithms, and supports efficient deployment using compact LLMs. Evaluated on multiple real-world customer service datasets, our method achieves performance on par with or superior to state-of-the-art contrastive learning–based approaches, while substantially reducing LLM inference overhead. It demonstrates strong scalability and adaptability under low-resource conditions.

Technology Category

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

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Economics, Online Markets and Human Computation: Humans versus LLMs for data annotation and labeling

Abstract

In this paper, we propose a training-free and label-free method for short text clustering that can be used on top of any existing embedder. In the context of customer-facing chatbots, companies are dealing with large amounts of user utterances that need to be clustered according to their intent. In these commercial settings, no labeled data is typically available, and the number of clusters is not known. Our method is based on iterative vector updating: it constructs sparse vectors based on representative texts, and then iteratively refines them through LLM guidance. Our method achieves comparable or superior results to state-of-the-art methods that use contrastive learning, but without assuming prior knowledge of clusters or labels. Experiments on diverse datasets and smaller LLMs show that our method is model agnostic and can be applied to any embedder, with relatively small LLMs, and different clustering methods. We also show that our method scales to large datasets, reducing the computational cost of the LLM. These low-resource, adaptable settings and the scalability of our method make it more aligned with real-world scenarios than existing clustering methods.

Problem

• Develops training-free clustering for short texts without labeled data
• Determines unknown intent clusters from user chatbot utterances
• Uses iterative vector updating with LLMs for scalable clustering

Innovation

• Training-free label-free clustering via iterative vector updates
• Sparse vector construction with LLM-guided iterative refinement
• Model-agnostic scalable method adaptable to any embedder