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 2304.07022cs.CL
收藏

标签依赖感知的集合预测网络用于多标签文本分类

Label Dependencies-aware Set Prediction Networks for Multi-label Text Classification

Sunshine Insurance Group Corporation Limited(阳光保险集团股份有限公司);Tsinghua University(清华大学)

Du Xinkai, Han Quanjie, Sun Yalin, Lv Chao, Sun Maosong

AI总结:

本文提出一种标签依赖感知的集合预测网络,利用图卷积网络建模标签相关性,并引入Bhattacharyya距离提升召回,在多标签文本分类任务上优于现有基线。

摘要:

多标签文本分类涉及从句子中提取所有相关标签。鉴于这些标签的无序性,我们提出将该问题视为一个集合预测任务。为了解决标签之间的相关性,我们利用图卷积网络,并基于标签之间的统计关系构建邻接矩阵。此外,我们通过将 Bhattacharyya 距离应用于集合预测网络的输出分布来增强召回能力。我们在两个多标签数据集上评估了我们的方法的有效性,并通过实验结果证明了其相对于先前基线的优越性。

英文摘要:

Multi-label text classification involves extracting all relevant labels from a sentence. Given the unordered nature of these labels, we propose approaching the problem as a set prediction task. To address the correlation between labels, we leverage Graph Convolutional Networks and construct an adjacency matrix based on the statistical relations between labels. Additionally, we enhance recall ability by applying the Bhattacharyya distance to the output distributions of the set prediction networks. We evaluate the effectiveness of our approach on two multi-label datasets and demonstrate its superiority over previous baselines through experimental results.

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。

↑