基于客观赋权法的多属性社交网络节点重要性排序

摘要

This paper studied node importance ranking of multi-attribute social network.The existing multi-attribute ranking was based on subjective determining method which may result in bias when researchers were not so familiar with the special so-cial field.In order to avoid the shortness,it suggested to use objective weight determining method—the entropy weighting meth-od when using un-familiar data.And it designed a ranking similarity to evaluate the similarity of tow ranking order.Simualtion compared the ranking similarity of different subjective weighting,subjective and objective weighting,and shows different subjec-tive weighting has lower similarity which proves the subjective weighting has different ranking order.And the objective weight has strong correlation with the subjective weighting which are familiar with the situation.So objective weighting can help users get reasonable ranking order.%研究多属性社交网络节点重要性排序问题。已有的多属性排序法基于主观赋权，在研究者对某数据集或者是某社交领域并不是很熟悉的情况下，主观赋权容易出现偏差。针对这点提出在对不熟悉的数据领域时采用客观赋权法—熵权法，进行权值的确定，为验证排序效果设计了排序相似度的计算方法。实例证明该相似度的合理性。实验对比不同主客观赋权排序结果的相似度，显示不同主观赋权表现出不同的相似度，证明主观判断存在差异，而客观赋权排序与熟悉情况下的主观赋权呈现强相关关系，能帮助使用者进行合理赋权。