摘要
针对传统方法在计算属性权重方面的不足,提出了一种与领域无关的基于Renyi熵的属性权重计算方法.基于概率论方法分析了属性的取值分布特征与属性权重之间的关系,并利用Renyi熵描述属性取值分布特征的合理性,最后得出基于Renyi熵的属性权重计算方法.该方法能从链接数据集中自动获取属性取值分布的Renyi熵,并自动计算出属性在共指分析中的权重.通过基于开放语义数据集的实验,以及与已有方法的结果对比,详细分析了该方法在属性权重计算方面的优势.
A domain-independent method of property weight calculation based on Renyi entropy was presented to make up for the deficiency of traditional methods. First the relation between property＇ s weight and its value distribution features of property was analyzed based on probability theory, and then the rationality of describing value distribution features with Renyi entropy was verified. Finally, the property weight calculation method based on Renyi entropy was proposed. This method can automatically obtain Renyi entropy of value distribution from the linked dataset and calculate the weight of property. The experiments based on the open semantic datasets and the comparative analyses with existing methods indicate that the Renyi entropy based method can obtain more reasonable weights of properties.
出处
《北京工业大学学报》
CAS
CSCD
北大核心
2014年第1期54-60,共7页
Journal of Beijing University of Technology
基金
教育部"211工程"公共服务体系建设重大专项(Q5007013201301)
