홈 › Papers

HCL-MTC Hierarchical Contrastive Learning for Multi-label Text Classification

Multi-label text classification is a big challenging subtask in text classification, where labels generally form a tree structure. Existing solutions learn the label tree structure in a shallow manner and ignore the distinctive information between labels. To address this problem, we propose a Hierarchical Contrastive Learning for Multi-label Text Classification (HCL-MTC), which constructs the graph based on the contrastive knowledge between labels. Specifically, we formulate the MTC as a multi-task learning by introducing a sampling hierarchical contrastive loss, which learns both the correlative and distinctive label information and is beneficial in learning deep label hierarchy. The experimental results show that the proposed model can achieve considerable improvements on both public datasets (i.e., RCV1-v2 and WoS).

Code (0)

등록된 구현이 없습니다.

Tasks

Methods 이 논문이 사용한 방법론

Similar Papers 제목 키워드 기반