用聚类感知方法提升大模型在图上的无标签学习可靠性
Where LLM Annotators Fail: Label-Free Learning on Graphs with LLMs
- 根据特征空间聚类估计大模型标注的局部可信度
- 在多个图数据集上显著优于现有无标签学习方法
- 适合处理大模型标注存在区域差异性噪声的任务
图节点分类通常需要标注节点，但在大规模图上获取标签成本高昂。当节点属性包含语义内容（如论文摘要、网页内容或产品描述）时，大语言模型（LLMs）可通过标注少量节点提供低成本监督。然而，这些由大模型生成的标签存在噪声，现有无标签学习方法通常将噪声视为全局或类别相关的。我们发现，大模型标注错误不仅与类别相关，还与区域相关：同一类别内，不同特征空间聚类的标注可靠性差异显著。为此，我们提出聚类感知噪声估计（CANE），一种无需真实标签的无标签学习框架，可估计聚类条件下的大模型可靠性，并据此决定哪些伪标签可信，哪些需修正。在多种图基准和GNN主干网络上，CANE均优于最强的无标签基线，尤其在具有更强聚类条件噪声的数据集上提升最明显。
原文摘要 · Abstract (English)
Node classification on graphs often requires labeled nodes, yet obtaining labels at graph scale is expensive. When node attributes contain semantic content, such as paper abstracts, web pages, or product descriptions, large language models (LLMs) can provide low-cost supervision by annotating a small subset of nodes. However, these LLM-generated labels are noisy, and existing label-free graph learning methods usually treat this noise as either global or class-conditional. We find that LLM annotation errors are not only class-dependent but also region-dependent: within the same class, reliability can vary sharply across feature-space clusters. In light of this, we propose Cluster-Aware Noise Estimation (CANE), a label-free learning framework that estimates cluster-conditional LLM reliability without ground truth labels, and uses this estimate to decide which pseudo-labels to trust, and which labels to correct. Across various graph benchmarks and GNN backbones, CANE improves over the strongest label-free baselines, with the largest gains on datasets exhibiting stronger cluster-conditional noise.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。