群智感知中采用节点社会属性的亲密度量化方法
Intimacy Relationship Quantification Method among Nodes Based on Social Attributes in Crowd Sensing

针对群智感知中节点社会关系计算存在的层次关系划分不清、关联因子描述粗糙等问题,提出一种采用社会属性的亲密度量化方法。该方法通过分析影响节点社会关系的多维要素,将节点属性因子划分为静态和动态两个维度。通过构造多维语义分级树和空间索引编码,对节点静态属性进行挖掘和形式化表示。同时,引入交互信息熵,对社会关系的不对称性进行分析和比较,以提高亲密度量化方法的客观性。基于层次分析法实现节点动态属性的二级评判和有效聚合,并通过节点融合度对亲密关系进行二次修正。实验结果表明:与已有方法相比,采用社会属性的亲密度量化方法预测准确率提高了14.67％,该方法能够有效降低群智感知中移动节点的误判概率,提高网络社团识别准确率,为候选服务节点集的选择提供有效依据。
A new method, called social？？based nodes intimacy quantification method (SNIQ), is proposed to solve the problem of unclear hierarchy and rough associated attribute description in nodes social relationships quantification. A formal representation of social attributes is presented by means of static semantic hierarchy tree and spatial index coding to analyze multi？？dimension factors that influence nodes social relationship. The asymmetry of social relations of nodes is reasonably evaluated by introducing the interactive information entropy. Furthermore, analytic hierarchy process (AHP) is used to achieve two？？level judgment and effective aggregation of node dynamic attributes, and a fusion degree of nodes is introduced to dynamically modify intimacy relationships of nodes. Experimental results show that SNIQ is better than the traditional algorithms in accuracy and recall rate in the process of getting nodes intimacy relationships, and effectively improves the success rate of nodes selection in crowd sensing. Comparisons with the existing algorithms show that the proposed SNIQ achieves 14.67% improvement in prediction accuracy