Label noise and long-tailed distributions are two major challenges in distantly supervised relation extraction. Recent studies have shown great progress on denoising, but pay little attention to the problem of long-tailed relations. In this paper, we introduce constraint graphs to model the dependencies between relation labels. On top of that, we further propose a novel constraint graph-based relation extraction framework(CGRE) to handle the two challenges simultaneously. CGRE employs graph convolution networks (GCNs) to propagate information from data-rich relation nodes to data-poor relation nodes, and thus boosts the representation learning of long-tailed relations. To further improve the noise immunity, a constraint-aware attention module is designed in CGRE to integrate the constraint information. Experimental results on a widely-used benchmark dataset indicate that our approach achieves significant improvements over the previous methods for both denoising and long-tailed relation extraction.
翻译:标签噪音和长尾分发是远程监控关系提取方面的两大挑战。最近的研究显示,在去除数据方面取得了很大进展,但很少注意长尾关系问题。在本文中,我们引入了制约图,以模拟关系标签之间的依赖关系。此外,我们进一步提议建立一个新的制约图基关系提取框架(CGRE ), 以同时应对这两个挑战。 国家地球科学研究与地球科学研究中心使用图变网络(GCNs ), 传播数据丰富关系节点与数据贫乏关系节点的信息, 从而推动长期关系的代表性学习。 为了进一步改善噪音豁免, 国家地球科学研究与地球科学研究中心设计了一个约束注意模块, 以整合限制关系信息。 广泛使用的基准数据集的实验结果显示,我们的方法大大改进了以往的脱色和长期关系提取方法。