Hierarchy-aware Label Semantics Matching Network for Hierarchical Text Classification

Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing, pages 4370–4379 August 1–6, 2021. ©2021 Association for Computational Linguistics 4370 Hierarchy-aware Label Semantics Matching Network for Hierarchical Text Classification Haibin Chen, Qianli Ma*, Zhenxi Lin, Jiangyue Yan School of Computer Science and Engineering, South China University of Technology, Guangzhou, China haibin chen@foxmail.com qianlima@scut.edu.cn∗ Abstract Hierarchical text classification is an important yet challenging task due to the complex struc- ture of the label hierarchy. Existing methods ignore the semantic relationship between text and labels, so they cannot make full use of the hierarchical information. To this end, we formulate the text-label semantics relationship as a semantic matching problem and thus pro- pose a hierarchy-aware label semantics match- ing network (HiMatch). First, we project text semantics and label semantics into a joint em- bedding space. We then introduce a joint em- bedding loss and a matching learning loss to model the matching relationship between the text semantics and the label semantics. Our model captures the text-label semantics match- ing relationship among coarse-grained labels and fine-grained labels in a hierarchy-aware manner. The experimental results on vari- ous benchmark datasets verify that our model achieves state-of-the-art results. 1 Introduction Hierarchical text classification (HTC) is widely used in Natural Language Processing (NLP), such as news categorization (Lewis et al., 2004) and sci- entific paper classification (Kowsari et al., 2017). HTC is a particular multi-label text classification problem, which introduces hierarchies to organize label structure. As depicted in Figure 1, HTC mod- els predict multiple labels in a given label hierarchy, which generally construct one or multiple paths from coarse-grained labels to fine-grained labels in a top-down manner (Aixin Sun and Ee-Peng Lim, 2001). Generally speaking, fine-grained labels are the most appropriate labels for describing the input text. Coarse-grained labels are generally the parent nodes of coarse- or fine-grained labels, expressing a more general concept. The key challenges of ∗*Corresponding author HTC are to model the large-scale, imbalanced, and structured label hierarchy (Mao et al., 2019). Root Economics Debt Revenue Society Coarse-grained Labels Fine-grained Labels Input Text: "Global debt is set to reach $200 trillion ..." Label Hierarchy Figure 1: An hierarchical text classification example tagged with labels Economics and Debt from coarse- grained label to fine-grained label. Existing work in HTC has introduced various methods to use hierarchical information in a holis- tic way. To capture the holistic label correlation features, some researchers proposed a hierarchy- aware global model to exploit the prior probabil- ity of label dependencies through Graph Convolu- tion Networks (GCN) and TreeLSTM (Zhou et al., 2020). Some researchers also introduced more la- bel correlation features such as label semantic sim- ilarity and label co-occurrence (Lu et al., 2020). They followed the traditional way to transform HTC into multiple binary classifiers for every label (F¨urnkranz et al., 2008). However, they ignored the interaction between text semantics and label se- mantics (F¨urnkranz et al., 2008; Wang et al., 2019), which …

Content truncated to 500 words