摘要
目的:面向真实世界的时序社会网络,实现基于其复杂关系不确定性的模糊化模型的建立,同时实现网络整体模糊化度量的时序预测。创新点:提出一种基于节点相似度的社会网络模糊化方法,并对网络模糊密度与模糊中心势进行预测,实现模糊网络的度量预测。方法:首先,考虑真实社会网络普遍存在的不确定性因素,提出一种基于网络节点相似度的模糊化方法,通过用模糊关系代替传统关系,可以将传统的社会网络转化为模糊的社会网络(图2)。然后,针对需要观测的网络密度及中心势两个整体度量(图1),同样根据网络的模糊化方法的定义可以得到网络模糊密度及模糊中心势。最后,结合模糊马尔可夫链模型,通过调整模糊度量的隶属函数以及模糊状态划分,来预测模糊度量变化并分析网络演化趋势。结论:针对真实的社会网络,提出一种社会网络的模糊化方法,实验说明了模糊网络可以比传统网络更好地描述网络演化过程。
Social network analysis (SNA) is among the hottest topics of current research. Most measurements of SNA methods are certainty oriented, while in reality, the uncertainties in relationships are widely spread to be overridden. In this paper, fuzzy concept is introduced to model the uncertainty, and a similarity metric is used to build a fuzzy relation model among individuals in the social network. The traditional social network is transformed into a fuzzy network by replacing the traditional relations with fuzzy relation and calculating the global fuzzy measure such as network density and centralization. Finally, the trend of fuzzy network evolution is analyzed and predicted with a fuzzy Markov chain. Experimental results demonstrate that the fuzzy network has more superiority than the traditional network in describing the network evolution process.
基金
Project supported by the National Natural Science Foundation of China(Nos.61272422 and 61202353)
关键词
时序网络
模糊网络
模糊马尔可夫链
Time-series network, Fuzzy network, Fuzzy Markov chain