作者：
Autodesk Research:
- Jian Zhao
- Michael Glueck
- Azam Khan
法国国家信息与自动化研究所(Inria):
- Fanny Chevalier
香港科技大学：
- Yanhong Wu
摘要
动态网络的以自我为中心的分析侧重于发现特定中心参与者(即自我网络)周围子网的时间模式。这些类型的分析在许多应用领域都很有用，例如社会科学和商业智能，提供了关于中心参与者如何与外部世界交互的见解。我们提出了EgoLines，一种支持动态网络自我中心分析的交互式可视化。使用一个“地铁地图”的比喻，用户可以在自我网络的演变中追踪一个单独的行动者。EgoLines的设计基于一系列与自我中心分析相关的关键分析问题，这些问题来自我们对三位领域专家的访谈和一般的网络分析任务。我们通过一个有 18 个参与者的对照实验和一个由领域专家开发的用例，证明了EgoLines在自我中心分析任务中的有效性。
Introduction
Dynamic Networks (macro)
Egocentric Analysis (micro)
Related Work
- Node-link diagrams (Hard to track changes)
- Matrices (Not suitable for sparse dynamic networks)
- Animation (High cognitive load)
- Projection (Missing topological information)
Analytical Questions
Dynamic ego-network analysis questions:
- Joining, leaving, and recurrence of a co-author?
- Connectivity to the focal author (ego)?
- Splitting and merging of co-author communities (clusters)?
- Stability of co-authorships
- …
Egolines
Inspiration: subway
author: line
Controlled study
- 18 participants
- 19 males and 5 females
- 13 analytical tasks (2 categories)
- temporal analysis, topological analysis
- 3 techniques
- EgoLines(EL), node-links(NL), small multiples(SM)
- 1 dataset
- IEEE VIS conferences co-authorship networks
Use-case Scenario
Domain expert evaluation
Limitations and future work
- More effective overview
- Reduce visual clutter
- Handle larger ego-networks
- Multi-scale aggregation of lines
- More experiments
- On other datasets and applications
- Shed light on facilitating the bridges-finding task