基于知识库的元类别医学期刊文章分类模型
Meta-Class Medical Journal Article Classification Based on Knowledge Base
韩春磊 1苏宇 2张玉志2
作者信息
- 1. 上海图书馆,上海 200135
- 2. 南开大学软件学院,天津 300457
- 折叠
摘要
医学期刊文章类别之间的相关性强,存在很多样本很少的类别,因此对自动分类工作造成困难.基于医学期刊分类的特点,设计了元类别的概念,以类别本身蕴含的信息作为分类的依据,通过外部知识库的信息支撑,提出了基于知识库的元类别医学期刊分类模型.通过和基准模型的比较,证明了本模型在类别相关性强、训练数据量低的医学期刊分类任务上具有明显的优势.
Abstract
The correlation among categories in medical journal articles is strong,and there are many categories with limited samples,making automatic classification challenging.Based on the characteristics of medical journal classification,the concept of meta-class is introduced,utilizing the inherent information of categories themselves as the basis for classification.With support from an external knowledge base for each class,a knowledge-based meta-class medical journal classification model is proposed.Through comparison with baseline models,it is demonstrated that this model exhibits significant advantages in medical journal article classification task with strong category correlations and limited training data.
关键词
文本分类/知识库/医学文本Key words
text classification/knowledge base/medical text引用本文复制引用
出版年
2024
NETL
NSTL
万方数据