Dongdaemun
SEOUL | KOREA
APRIL 18-21, 2017
News
Sponsors
Previous Events
Channels
Logo
You can download the logo here.

GraphLDA: Latent Dirichlet Allocation-based Visual Exploration of Dynamic Graphs

  • Lijing Lin
    Peking University, Beijing, China
  • Siming Chen
    Peking University, Beijing, China
  • Fan Hong
    Peking University, Beijing, China
  • Chufan Lai
    Peking University, Beijing, China
  • Shuai Chen
    Peking University, Beijing, China
  • Xiaoru Yuan
    Peking University, Beijing, China

Abstract

In dynamic graph visualization and analysis, it is very challenging to visualize both the overall evolution of trends and the detailed changes of structures simultaneously. In this work, we propose a latent Dirichlet allocation (LDA) -based visual exploration method for dynamic graphs. With the LDA-based analysis, we can reveal important structures in the dynamic graph based on the extracted se- mantic topics. To gain a deeper understanding of the derived structures and their evolution, we propose a visual analytics pipeline enabling users to interpret and explore the dynamic graph. To experiment with the proposed method, we provide a visual analytics system to test with real-world data. Our case on the datasets of dynamic collaboration network has demonstrated the effectiveness of the proposed method.