多搜索引擎权重计算及搜索结果排序质量评估  

Weight calculation for search engines and quality evaluation for ranking of search results

作　　者：李超[1,2,3] 谢坤武[1,2] 

机构地区：[1]湖北民族学院计算机科学与技术系,湖北恩施445000 [2]湖北民族学院科技学院,湖北恩施445000 [3]河海大学计算机与信息学院,南京211000

出　　处：《计算机工程与应用》2014年第12期21-25,共5页Computer Engineering and Applications

基　　金：国家自然科学基金专项基金(No.61362012);湖北省自然科学基金(No.2009CDB069);湖北民族学院科技学院项目(No.K201001)

摘　　要：搜索引擎在多成员搜索引擎搜索结果的整合过程中,搜索结果的排序在很大程度上决定着元搜索引擎的服务质量。为了实现搜索结果的有效整合,目前技术主要结合查询请求、文档内容、初始排序或(和)赋予搜索成员搜索引擎权重等因素。其中采用赋予搜索引擎权重时,往往根据用户和技术人员经验,主观地进行赋值,不能体现真实的用户搜索偏好。为此,提出了通过挖掘用户搜索及遍历情况,动态地赋予各成员搜索引擎权重的方法。通过用户遍历及点击下载情况,得到了用户搜索遍历与返回结果的匹配度,论证了该方法的可行性和有效性。In the integration process of search results returned from multi member search engines, the quality of a service meta-search engine is determined by the ranking of search results to a great extent. The current technologies on achieving effective integration of search results mainly rely on combining search queries, document content, the initial sorting infor-mation or（and）endowing the member search engines with weights, and others. The method of endowing engines with weights is objective, often depends on the subjective experience of users, and can’t embody users’search preferences. So, this paper proposes a method to dynamically endow the member search engines with weights based on mining user’s search-ing and navigation habits. It analyzes users’log data on clicking and downloading search results, and gets the matching degrees between search results and users’navigation. The statistic results demonstrate that the methodology is feasible and effective to the ranking of multi engine search results.

关 键 词：搜索引擎 权重计算 排序 质量评估 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]