Visualization strategies facilitate streaming network analysis by allowing its exploration through graphical and interactive layouts. Depending on the strategy and the network density, such layouts may suffer from a high level of visual clutter that hides meaningful temporal patterns, highly active groups of nodes, bursts of activity, and other important network properties. Edge sampling improves layout readability, highlighting important properties and leading to easier and faster pattern identification and decision making. This paper presents Streaming Edge Sampling for Network Visualization–SEVis, a streaming edge sampling method that discards edges of low-active nodes while preserving a distribution of edge counts that is similar to the original network. It can be applied to a variety of layouts to enhance streaming network analyses. We evaluated SEVis performance using synthetic and real-world networks through quantitative and visual analyses. The results indicate a higher performance of SEVis for clutter reduction and pattern identification when compared with other sampling methods.

可视化策略通过允许其通过图形和交互式布局进行浏览来促进流网络分析。根据策略和网络密度的不同，此类布局可能会遭受高度的视觉混乱，这些视觉混乱会隐藏有意义的时间模式，高度活跃的节点组，活动爆发以及其他重要的网络属性。边缘采样提高了版面的可读性，突出了重要的特性，并使得图案识别和决策更加容易，快捷。本文提出了用于网络可视化的流边缘采样–SEVis，一种流式边缘采样方法，可丢弃低活动节点的边缘，同时保留与原始网络相似的边缘计数分布。它可以应用于各种布局，以增强流网络分析。通过定量和视觉分析，我们使用综合和现实网络评估了SEVis的性能。结果表明，与其他采样方法相比，SEVis在减少杂波和模式识别方面具有更高的性能。