摘要
社团结构作为社会网络的一个重要特征,对社会网络的可视化结果应当尽可能真实地反映在高维空间中节点间的距离,进而体现社会网络的聚类情况。针对传统力引导算法存在的无法展示社会网络中社团结构信息的弊端,提出一种聚类效果更加突出的改进布局算法。首先,改进斥力受力公式,引入乘积因子;然后,基于节点间相似度定义增量相似度,再分别定义两个单调函数将增量相似度映射为弹簧原长与新增的斥力乘积因子,实现节点相似度信息的嵌入;最终,在不对社会网络提前进行社团划分的前提下仅通过布局便能展示其中的社团结构。实验结果表明,所提算法与传统力引导算法相比,在展示社会网络的聚簇性方面性能领先。
Community structure is an important feature of social networks. The visualization of social networks should reflect the distance between vertexes in high dimensional space as realistic as possible, and thus reflect the clustering of social networks. Aiming at the shortcomings of traditional force-directed algorithm which cannot show the information of community structure in social network, a more efficient layout algorithm with clustering effect was proposed. Firstly, the repulsive force formula was improved by introducing a multiplication factor. Next the incremental similarity based on the similarity between vertexes was defined, and then two monotone functions were defined that mapping the incremental similarity to the original length of the spring and the new repulsion multiplication factor. The improved algorithm shows the community structure without dividing a social network into communities in advance. The experimental results show that the proposed algorithm is better than the traditional force-directed algorithm in demonstrating the clustering of social networks.
出处
《计算机应用》
CSCD
北大核心
2017年第A02期214-218,224,共6页
journal of Computer Applications
基金
国家自然科学基金资助项目(61503312)
关键词
数据可视化
社会网络
社团结构
相似度
力引导算法
data visualization
social network
community structure
similarity
force-directed algorithm
