网络信息检索 文档检索系统评价.pptVIP

7/7/2005 7/7/2005 Lecture 1 Overview Lecture 1 Overview Lecture 2 Evaluation in Document Retrieval 文档检索系统评价 Reference: James Allan, University of Massachusetts Amherst Review We had proposed a very simple model of document retrieval Recall that IR is much more than document retrieval! Statistical language model（统计语言模型） Rank documents by P(Q|Di) Assume independence so Estimate P(qi|D) by tf(w,D) / len(D) Highly artificial examples suggested model is “okay” Is it really any good? How can we find out? (Do you think Google better than Baidu or vice versa? How can you prove that?) How ca