PRIS信息检索技术报告

PRIS Information Retrieval System Report

建立索引、查询扩展和相似度计算等都是检索系统中的关键环节。Pills信息检索系统更侧重于构造简单有效的查询扩展算法。本报告介绍了北京邮电大学模式识别实验室参加2005年863信息检索测试的系统结构和具体方法。本报告分别介绍了预处理、分词、建立索引、查询扩展和相似度计算等部分。最后针对测试结果进行了分析。对正式评测的50个主题粜检索，获得的三项评价指标为：程序自动构造查询时，MAPm-0．1862，P@10=0．5180，R．Precision=0．2554；人工构造查询时，MAP=0．1862，P@10=0．5180，R-Precision=0．2554。

Index, query expansion and similarity computation are all the critical steps in information retrieval. More attention is paid to query expansion in constructing PRIS information retrieval system. The main content of this report is the introduction to the system which is designed by PRIS for 2005 863 information retrieval test. The structure of the system and the key means are introduced. And the result analysis is also presented. In the official evaluation on 50 topics, MAP 0. 1862, P@ 10 0. 5180, R-Precision 0. 2554 and MAP 0. 1862, P@ 10 0. 5180, R-Precision 0. 2554 are achieved with queries constructed automatically and artificially respectively.