基于语义网高效信息查询方法.docVIP

基于语义网高效信息查询方法 　　摘要：为了提高Web信息检索的准确率，提出一种基于语义网的高效信息查询方法。首先从本体库中提取目标资源与查询关键字之间的语义路径，通过分析语义路径所包含的属性的权重和识别能力，分别计算每个语义路径的权重；然后，根据资源与查询关键字之间的语义路径的权重、数量和特异性，分别计算每个资源与各关键字之间的语义相关性，并结合关键字的涵盖范围和识别能力综合计算每个资源与关键字集之间的语义相关性；最后，以该相关性为依据对所有资源进行排序和输出。实验结果表明，与OntoLook、tf*idf和TMSubtree三种语义网查询算法相比，基于语义网的高效信息查询方法的平均正确率分别提高了69.0、25.0和21.0个百分点；平均召回率分别提高了77.1、28.3和24.3个百分点；平均F测度值分别提高了72.4、26.4和22.4个百分点。实验结果表明：该方法不仅能够有效提升语义查询的准确率，而且对隐性信息也有很好的查询效果。 　　关键词：语义网；本体；语义关系路径；权重计算；特异性 　　中图分类号： TP391.3 　　文献标志码：A 　　Abstract： In order to improve the accuracy of Web information retrieval， an efficient information search method based on semantic Web was proposed. Firstly， all semantic paths between the target resources and the query keywords were extracted from the ontology library， and the weight of each semantic path was calculated by analyzing the weight and identification power of attributes included in it. Then， based on the weights， the number and the specificity of the semantic paths between resources and query keywords， as well as the semantic correlation between each resource and each keyword were calculated； and combining with the coverage and identification power of each keyword， the semantic correlation between each resource and the keyword set was calculated. Finally， on the basis of the correlation， all the resources were sorted and output. The experimental results show that compared with three different semantic Web search algorithms， including OntoLook， tf*idf and TMSubtree， the proposed method improved the average precision of 69.0， 25.0， 21.0 percentage points， respectively； average recall of 77.1， 28.3， 24.3 percentage points， respectively； and average Fmeasure of 72.4， 26.4， 22.4 percentage points， respectively. These results prove the proposed method can not only effectively improve the accuracy of semantic search， but also have good query results for indirect information. 　　Key words： semantic Web； ontology； semantic relationship path； weight calculation； specificity 　　0引言