首页 | 本学科首页   官方微博 | 高级检索  
     检索      

文本分类的性能评估指标
引用本文:张启蕊,董守斌,张凌.文本分类的性能评估指标[J].广西师范大学学报(自然科学版),2007,25(2):119-122.
作者姓名:张启蕊  董守斌  张凌
作者单位:华南理工大学,广东省计算机网络重点实验室,广东,广州,510640
摘    要:在信息检索领域,查全率与查准率是一对相互制约的指标.为了研究文本分类领域查全率和查准率的关系,在此从理论和实验两方面分析查全率及测试集对查准率的影响.理论分析与实验结果一致得出,在文本分类中查全率和查准率是两个一致的指标.另外,在查全率确定的情况下,测试集中各类别文档比例的变化也会导致查准率的变化.

关 键 词:文本分类  查准率  查全率  测试集
文章编号:1001-6600(2007)02-0119-04
收稿时间:2006-12-28
修稿时间:2006-12-28

Performance Evaluation in Text Classification
ZHANG Qi-rui,DONG Shou-bin,ZHANG Ling.Performance Evaluation in Text Classification[J].Journal of Guangxi Normal University(Natural Science Edition),2007,25(2):119-122.
Authors:ZHANG Qi-rui  DONG Shou-bin  ZHANG Ling
Institution:Guangdong Key Laboratory of Computer Network,South China University of Technology,Guangzhou 510640,China
Abstract:In information retrieval, recall and precision are regarded as two inter-constraint measures. In order to discuss the relation of recall and precision in text categorization ,the impact of recall and test sets on precision is analyzed through theory and experiment respectively. In text categorization, recall and precision are consistent. In addition,if recall is confirmed ,precision is still influenced by the portion of documents in test sets.
Keywords:text categorization  precision  recall  test set
本文献已被 CNKI 维普 万方数据 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号