高级检索

空间位置耦合的地理社交网络可视化布局算法

Geographical Social Network Visualization Layout Algorithm Based on Spatial Location Coupling

  • 摘要: 针对传统力引导布局算法无法兼顾节点初始地理空间位置特征的问题,提出了空间位置耦合的力引导算法(SCFDA),该算法在节点布局时,使节点除了受到胡克引力和库伦斥力影响外,还受到来自节点隶属的空间社团的中心重力和边界斥力影响,这样节点将在一定地理空间范围约束下实现布局和位置调整.在计算中心重力时顾及节点的内部度因素,使得内部度越高的节点越靠近空间社团的中心,在计算边界斥力时兼顾节点的外部度因素,使得外部度越高的节点越靠近空间社团的边界.利用2组社交网络签到数据集Gowalla和Brightkite进行了实验,通过兼顾内部度和外部度因素的综合评价指标值E(G)对实验结果进行评价,横向对比实验结果表明,SCFDA的E(G)值约为传统力引导布局算法的十分之一,而E(G)值越小则代表布局结果在顾及节点空间位置特征方面越合理;纵向对比实验结果表明,SCFDA在不同数据集和不同数据量上均具有普适性.

     

    Abstract: In view of the problem that the traditional force-guided layout algorithm could not take into account the initial geospatial location characteristics of the node,this paper proposes the spatially coupled force-directed algorithm(SCFDA).While during layout,the node is affected by the Hulk gravity and the Coulomb repulsion,and also receives the boundary repulsion and central gravity from the spatial community it belongs to.As consequence,the node will adjust the layout and position under certain geographical space constraints.When calculating the center gravity,the internal degree of the node is taken into account,so that the nodes with higher internal degree are closer to the center of the spatial community.In calculating the boundary repulsion,the external degree of the node is taken into account,so that the nodes with higher external degree are closer to the boundary of the spatial community.Gowalla and Brightkite,the two of social network check-in datasets are used to conduct experiments with the consideration of internal degree and external degree.The comprehensive evaluation index value E(G)evaluates the results of the experiments.The horizontal comparison experiment result shows that the SCFDA E(G)value is about one-tenth of the traditional force-directed layout algorithm,which indicates more reasonable layout result when taking into account the spatial characteristics of the nodes with smaller value.The vertical comparison experiment result shows that the SCFDA is universal on different data sets and different data volumes.

     

/

返回文章
返回