一种关联多本体的科技奖励检索方法  

A Science and Technology Award Retrieval Method of Associating Multiple-Ontologies

作　　者：曾辉[1] 江健健[1] 熊李艳[1] 黄卫春[1] 

机构地区：[1]华东交通大学信息工程学院,江西南昌330013

出　　处：《华东交通大学学报》2014年第2期112-118,共7页Journal of East China Jiaotong University

基　　金：国家自然科学基金项目(61363072)

摘　　要：现有的科技奖励检索都是基于关键词的匹配,忽略了对用户查询的语义理解。针对传统信息检索存在的问题以及结合当前面向实际应用的基于本体的语义检索的特点,提出一种关联多本体的科技奖励检索方法。通过对科技奖励项目的领域本体构建,对概念相似度计算方法的改进及关联多本体概念,使扩展词语更能表达用户检索意图。实验结果表明,该方法对比传统查询检索和单独本体扩展检索获得了更好的召回率和准确率。The existing science and technology award retrieval is based on keywords matching technology, which ignores the semantic understanding of user queries. To solve the problem of traditional information retrieval and combine characteristics of current practical application-oriented and ontology-based semantic retrieval, this pa-per proposes a science and technology award retrieval method of associating multiple-ontologies. By constructing the domain ontology for science and technology awards projects and by improving the concept similarity calcula-tion method and associating multiple-ontologies concepts, the proposed retrieval method makes the query expan-sion words further express retrieval intention of users. The experimental results show the method obtains better pre-cision and recall compared with keywords query and single-ontology extensions query.

关 键 词：科技奖励 关联多本体 本体构建 概念相似性 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]