Continual Graph Convolutional Network for Text Classification


  • Tiandeng Wu Huawei Technolologies Co., Ltd
  • Qijiong Liu The Hong Kong Polytechnic University
  • Yi Cao Huawei Technolologies Co., Ltd
  • Yao Huang Huawei Technolologies Co., Ltd
  • Xiao-Ming Wu The Hong Kong Polytechnic University
  • Jiandong Ding Huawei Technolologies Co., Ltd



SNLP: Text Classification, SNLP: Information Extraction, SNLP: Learning & Optimization for SNLP, SNLP: Ontology Induction From Text, SNLP: Sentence-Level Semantics and Textual Inference


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




How to Cite

Wu, T., Liu, Q., Cao, Y., Huang, Y., Wu, X.-M., & Ding, J. (2023). Continual Graph Convolutional Network for Text Classification. Proceedings of the AAAI Conference on Artificial Intelligence, 37(11), 13754-13762.



AAAI Technical Track on Speech & Natural Language Processing