Dual Refinement Cycle Learning: Unsupervised Text Classification of Mamba and Community Detection on Text Attributed Graph

2025-12-07

AI Summary

To address the performance degradation in community detection and text classification on text-attributed graphs caused by label scarcity, this paper proposes a structural-semantic dual-refinement cyclic learning framework. The method introduces a novel bidirectional co-optimization mechanism between a GCN-based Community Detection Module (GCN-CDM) and a Text Semantic Modeling Module (TSMM), enabling unsupervised joint enhancement of graph structure and textual semantics via iterative pseudo-labeling. Furthermore, it integrates community signals into the Mamba architecture to construct the first annotation-free, graph-guided generative text classifier. Evaluated on multiple benchmark datasets, the approach significantly improves both structural cohesion and semantic consistency of detected communities. Remarkably, the Mamba classifier trained solely on community signals achieves accuracy comparable to fully supervised baselines, effectively bridging the gap between unsupervised graph representation learning and downstream text understanding.

Technology Category

• Machine Learning: Graph-based Machine Learning
• Data Mining & Knowledge Management: Graph Mining, Social Network Analysis & Community
• Natural Language Processing: Sentence-level Semantics, Textual Inference, etc.

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Web Mining and Content Analysis: Bridging structured and unstructured data

Abstract

Pretrained language models offer strong text understanding capabilities but remain difficult to deploy in real-world text-attributed networks due to their heavy dependence on labeled data. Meanwhile, community detection methods typically ignore textual semantics, limiting their usefulness in downstream applications such as content organization, recommendation, and risk monitoring. To overcome these limitations, we present Dual Refinement Cycle Learning (DRCL), a fully unsupervised framework designed for practical scenarios where no labels or category definitions are available. DRCL integrates structural and semantic information through a warm-start initialization and a bidirectional refinement cycle between a GCN-based Community Detection Module (GCN-CDM) and a Text Semantic Modeling Module (TSMM). The two modules iteratively exchange pseudo-labels, allowing semantic cues to enhance structural clustering and structural patterns to guide text representation learning without manual supervision. Across several text-attributed graph datasets, DRCL consistently improves the structural and semantic quality of discovered communities. Moreover, a Mamba-based classifier trained solely from DRCL's community signals achieves accuracy comparable to supervised models, demonstrating its potential for deployment in large-scale systems where labeled data are scarce or costly.

Problem

• Unsupervised text classification on text-attributed graphs without labels
• Integrating structural and semantic information for community detection
• Enhancing text representation learning using structural patterns automatically

Innovation

• Unsupervised framework integrates structural and semantic information
• Iterative bidirectional refinement cycle exchanges pseudo-labels between modules
• Mamba-based classifier trained from community signals achieves supervised accuracy