arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2304.04152cs.CL
收藏

用于文本分类的持续图卷积网络

Continual Graph Convolutional Network for Text Classification

Huawei Technologies Co., Ltd.(华为技术有限公司);The Hong Kong Polytechnic University(香港理工大学)

Tiandeng Wu, Qijiong Liu, Yi Cao, Yao Huang, Xiao-Ming Wu, Jiandong Ding

AI总结:

提出持续图卷积网络ContGCN,通过全token-任意文档范式动态更新文档-token图,结合出现记忆模块和自监督对比学习,实现无标签在线流式文本推理,在华为舆情系统A/B测试中性能提升8.86%。

摘要:

图卷积网络(GCN)已被成功应用于捕获文本分类的全局非连续和长距离语义信息。然而,尽管基于GCN的方法在离线评估中显示出有希望的结果,但它们通常遵循已见token-已见文档范式,通过构建固定的文档-token图,无法对新文档进行推理。将它们部署在在线系统中以推断流式文本数据是一个挑战。在这项工作中,我们提出了一个持续GCN模型(ContGCN),将推理从已观察文档推广到未观察文档。具体而言,我们提出了一种新的全token-任意文档范式,在在线系统的训练和测试阶段的每个批次中动态更新文档-token图。此外,我们设计了一个出现记忆模块和一个自监督对比学习目标,以无标签方式更新ContGCN。在华为舆情分析系统上进行的为期3个月的A/B测试表明,与最先进的方法相比,ContGCN实现了8.86%的性能提升。在五个公共数据集上的离线实验也显示ContGCN能提高推理质量。源代码将在https://github.com/Jyonn/ContGCN发布。

英文摘要:

Graph convolutional network (GCN) has been successfully applied to capture global non-consecutive and long-distance semantic information for text classification. However, while GCN-based methods have shown promising results in offline evaluations, they commonly follow a seen-token-seen-document paradigm by constructing a fixed document-token graph and cannot make inferences on new documents. It is a challenge to deploy them in online systems to infer steaming text data. In this work, we present a continual GCN model (ContGCN) to generalize inferences from observed documents to unobserved documents. Concretely, we propose a new all-token-any-document paradigm to dynamically update the document-token graph in every batch during both the training and testing phases of an online system. Moreover, we design an occurrence memory module and a self-supervised contrastive learning objective to update ContGCN in a label-free manner. A 3-month A/B test on Huawei public opinion analysis system shows ContGCN achieves 8.86% performance gain compared with state-of-the-art methods. Offline experiments on five public datasets also show ContGCN can improve inference quality. The source code will be released at https://github.com/Jyonn/ContGCN.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑