中文摘要：

文章结合蒙古文的独特性,研究蒙古文信息检索系统。首先搭建一个用于评价检索性能的蒙古文文档测试集,建立一套蒙古文信息检索系统。实验对比分析检索模型、平滑算法、蒙古文停用词表、词干还原和伪相关反馈等技术对蒙古文信息检索系统关键技术对检索性能的影响。实验结果表明,蒙古文信息检索系统选择结构化语言模型、Dirichlet平滑方法、停用词表、以词根做检索单元和伪相关反馈可以更好地提升检索性能。

英文摘要：

This paper focuses on Mongolian information retrieval （IR） . The authors build a standard Mongolian document dataset for evaluating the IR performance and set up a Mongolian information retrieval system. Based on them, the influences of related technologies such as IR models, smoothing algorithms, Mongolian stop words, Mongolian word stemming and pseudo-relevance feedback, are compared and analyzed. Experimental results show that the structured language model, Dirichlet smoothing, stop words, word stemming and pseudo-relevance feedback can improve the performance of the Mongolian information retrieval system.