Large-scale multi-label text classification (LMTC) aims to associate a document with its relevant labels from a large candidate set. Most existing LMTC approaches rely on massive human-annotated training data, which are often costly to obtain and suffer from a long-tailed label distribution (i.e., many labels occur only a few times in the training set). In this paper, we study LMTC under the zero-shot setting, which does not require any annotated documents with labels and only relies on label surface names and descriptions. To train a classifier that calculates the similarity score between a document and a label, we propose a novel metadata-induced contrastive learning (MICoL) method. Different from previous text-based contrastive learning techniques, MICoL exploits document metadata (e.g., authors, venues, and references of research papers), which are widely available on the Web, to derive similar document-document pairs. Experimental results on two large-scale datasets show that: (1) MICoL significantly outperforms strong zero-shot text classification and contrastive learning baselines; (2) MICoL is on par with the state-of-the-art supervised metadata-aware LMTC method trained on 10K-200K labeled documents; and (3) MICoL tends to predict more infrequent labels than supervised methods, thus alleviates the deteriorated performance on long-tailed labels.
翻译:大型多标签文本分类(LMTC)的目的是将文件与其来自大型候选数据集的相关标签联系起来。大多数现有的LMTC方法都依赖大量人文附加说明的培训数据,这些数据往往要花费昂贵才能获得并受到长尾标签分发(例如,许多标签只在培训组中出现过几次)的影响。在本文中,我们在零点设置下研究LMTC,它不需要带有标签的任何附加说明的文件,而只依赖标签表面名称和描述。为了培训一个计算文件和标签之间相似度分的分类员,我们建议采用新的由元数据引发的对比学习(MICOL)方法。不同于以往基于文本的对比学习技术,MICOL利用文件元数据(例如,作者、地点和参考研究文件),在网上广泛提供类似的文档配对。两个大型数据集的实验结果显示:(1) MICol 明显超越了强的零点文本分类和对比学习基线;(2) MIC-MT-S-C-Sqrentral-rolex-lax-lax-lax-s lax-tracis lax-tracal-rmal-reval-rodal-leval-leval-lation-lation-leval-leval-lation-lation-lational-lational-lation-lational-lational-lational-lation-l-l-l-l-lation-lation-s-s-s-lation-lation-lation-s-s-lxxxxx-lation-lation-lation-lation-l-l-lation-lation-l-l-l-lxxxx-s-s-s-s-s-s-s-lxxxxxxxx-s-s-s-s-lxxxxxxxx-s-s-s-lxxxxxx-s-s-s-s-s-s-s-l-l-l-l-l-l-l-l-l-l-l-l-l-l-l-l-l-l-l-l-l-l-l-l-l-l-l-l-l-