摘要
查全率,查准率的高低在很大程度上决定了一个智能搜索引擎的性能,而用户偏好库中的算法如果运用得当可明显提高查全率和查准率.在探讨了在基于本体的智能搜索中用户偏好库的作用和其类型,特别是客观世界中某一特定领域或主题的用户偏好库之后,详细地研究了用户偏好库中用户兴趣剖像生成的提取算法,TF*IDF算法和TF*IWF*IWF算法和基于本体的查询扩展算法,并且讨论了各个算法之间的联系.论述了用户偏好库在基于本体的搜索系统中的独特意义.通过实验最终验证了这些算法可达到查全率为10.1%、查准率为81%的良好效果,进而证实基于本体在智能搜索中所体现的优势.
The level of precision and recall largely determines a smart search engine performance. If user preferences algorithm is used properly, it can significantly enhance the recall ratio and precision ratio. After study on the user＇s preference storage＇s function and type based on the ontology＇s intelligent search in some particular fields or themes, this article researchs user＇s preference＇s information and its extraction algorithm- TF ~＂ IDF algorithm, TF ~ IWF ~ IWF algorithm and inquiry expansion algorithm based on ontology. And we discuss the relations among those algorithms and the unique significance of user＇s preference in the search system based on ontology. Finally experiments confirm these algorithms to achieve the recall of ratio 10. 1% and precision of ratio 81%.
出处
《西安工业大学学报》
CAS
2007年第5期480-484,共5页
Journal of Xi’an Technological University
基金
国家"星火计划"项目--"西北农业专家远程信息化服务体系示范"(2004EA850069)
