摘要
信息检索系统的效果评价是信息检索研究必不可少的环节,也是检验信息检索理论和进行方法研究的重要途径之一。本文首先从相关性理论出发,讨论了信息检索中检出文档相关性的维度和度量问题,以及等级相关性判断和大规模数据集的相关性评估方法。其次,进一步介绍了信息检索系统效果评价的基本指标查全率和查准率,并对常用评价指标MAP、Bpref等进行了比较和分析。最后,笔者认为未来信息检索系统效果评价的研究应注重将用户纳入评价体系中,使评价结果更具有真实性和实用性。
The evaluation on the effectiveness of information retrieval system is indispensable in information retrieval study, and is also one of the important approaches to testify theories and study methods in IR fields. Firstly, this paper discusses the obstacles about dimensions and measures of relevance, and approches for judging relevance of documents on the basis of relevance framework, such as rank relevance judgment, and etc. Secondly, the paper focuses on the two basic measures （Precision and Recall ） , and frequently used measures for evaluating and IR system, then make a comparison between them. Finally, the future study of information retrieval effectiveness evaluation should give more focus on Users, which makes the evaluation results more valuable.
出处
《图书情报知识》
CSSCI
北大核心
2009年第4期89-94,共6页
Documentation,Information & Knowledge
基金
国家社科基金青年项目"基于网络信息计量的网络信息资源配置与调控机制研究"(05CTQ004)研究成果之一
关键词
信息检索
相关性判断
评价
Information retrieval
Relevance judgment
Evaluation
作者简介
秦喜艳，女，硕士生；
陆伟，男，博士、副教授；
姜捷璞，男，硕士生。
