基于文档重排列的中文信息检索系统(英文)

Using Document Re-ranking for Chinese Information Retrieval System

对于文本信息检索,用户都希望从被检索出来的前N篇文章中得到更多的相关信息。本文介绍一个基于文档重排列的中文信息检索系统。为了通过重排列初检索文本来提高检索结果的精确率,该系统按照初检索结果中前100个排列文档中的关键词的分布对结果中的所有1000个文档进行重排列。实验中使用NTCIR-3正式的中文测试数据作为测试集,结果表明,该系统对中文文本检索精确率的提高取得一定的效果。

For information retrieval, users hope to acquire more relevant information from the top N ranking documents. In this paper, we describe a Chinese Information Retrieval (IR) system that use a document re-ranking method based on term distribution of the top 100 ranking documents, to improve the precision of the 1000 ranking documents by reordering the initial retrieval documents. The experiment with NTCIR-3 formal Chinese test collection shows that the proposed method of document re-ranking can improve the precision of the retrieved documents.