IOCC: Aligning Semantic and Cluster Centers for Few-shot Short Text Clustering

TL;DR

This paper introduces IOCC, a novel few-shot contrastive learning approach that aligns cluster centers with semantic centers in short text clustering, significantly improving performance on benchmark datasets.

Contribution

The paper proposes IOCC, combining interaction-enhanced optimal transport and center-aware contrastive learning to better capture semantic structures in short text clustering.

Findings

Outperforms previous methods on eight benchmark datasets.

Achieves up to 7.34% improvement on Biomedical dataset.

Enhances clustering stability and efficiency.

Abstract

In clustering tasks, it is essential to structure the feature space into clear, well-separated distributions. However, because short text representations have limited expressiveness, conventional methods struggle to identify cluster centers that truly capture each category's underlying semantics, causing the representations to be optimized in suboptimal directions. To address this issue, we propose IOCC, a novel few-shot contrastive learning method that achieves alignment between the cluster centers and the semantic centers. IOCC consists of two key modules: Interaction-enhanced Optimal Transport (IEOT) and Center-aware Contrastive Learning (CACL). Specifically, IEOT incorporates semantic interactions between individual samples into the conventional optimal transport problem, and generate pseudo-labels. Based on these pseudo-labels, we aggregate high-confidence samples to construct pseudo-centers that approximate the semantic centers. Next, CACL optimizes text representations toward their corresponding pseudo-centers. As training progresses, the collaboration between the two modules gradually reduces the gap between cluster centers and semantic centers. Therefore, the model will learn a high-quality distribution, improving clustering performance. Extensive experiments on eight benchmark datasets show that IOCC outperforms previous methods, achieving up to 7.34\% improvement on challenging Biomedical dataset and also excelling in clustering stability and efficiency. The code is available at: https://anonymous.4open.science/r/IOCC-C438.