计算机科学与探索 ›› 2015, Vol. 9 ›› Issue (5): 575-585.DOI: 10.3778/j.issn.1673-9418.1409008

• 系统软件与软件工程 • 上一篇    下一篇

Web服务描述的本体学习方法

田  刚1,2+,何克清2,孙承爱1,赵卫东1,赵  一2   

  1. 1. 山东科技大学 信息科学与工程学院,山东 青岛 266590
    2. 武汉大学 计算机学院 软件工程国家重点实验室,武汉 430072
  • 出版日期:2015-05-01 发布日期:2015-05-06

Ontology Learning from Web Service Descriptions

TIAN Gang1,2+, HE Keqing2, SUN Cheng’ai1, ZHAO Weidong1, ZHAO Yi2   

  1. 1. College of Information Science and Engineering, Shandong University of Science and Technology, Qingdao, Shandong 266590, China
    2. State Key Laboratory of Software Engineering, School of Computer, Wuhan University, Wuhan 430072, China
  • Online:2015-05-01 Published:2015-05-06

摘要: 目前互联网上已有的本体难以满足Web服务语义查询的需要,而手工建立本体不仅困难而且成本很高,因此有必要建立一种从已有Web服务描述中进行本体学习的方法,辅助领域专家建立高质量的领域本体。针对上述问题,提出了一种针对Web服务描述的本体学习方法。该方法利用一种基于层次Dirichlet过程(hierarchical Dirichlet process,HDP)的主题模型自动学习本体层次结构和每一层中所包含的主题数目。每一层次的主题采用“代表单词”表示,“代表单词”由算法计算得出。基于参数组合模式的规则定义语义丰富规则,并被应用在自底向上的本体语义丰富算法中。实验表明,该方法在语义内容上要比单独使用hHDP(hierarchies of hierarchical Dirichlet process)方法更加丰富,在语义层次上要好于使用关联规则挖掘方法形成的本体。

关键词: Web服务参数, 本体学习, 层次Dirichlet过程(HDP)

Abstract: At present, ontologies on the Internet are hard to satisfy the needs of semantic searching on Web services. Manually building ontologies referring to specific application is difficult and costly, so that it is necessary to establish a method of ontology automatically learning from Web service descriptions to facilitate domain experts generating high quality ontologies. In view of these problems, this paper proposes an ontology learning method from Web service descriptions. This method automatically learns ontology hierarchical structures and topics in each level by using topic model based on HDP (hierarchical Dirichlet process). Topics in each level are represented by “representation word” whose calculation is defined. Rules according to parameters composition pattern which define semantic enriching rules are utilized in the bottom-up ontology sematic enriching algorithm. Experiments show that the proposed method is richer in semantics than hHDP (hierarchies of hierarchical Dirichlet process) and better in semantic hierarchies than the method using association rule mining.

Key words: Web service parameter, ontology learning, hierarchical Dirichlet process (HDP)