Understanding Evolution Patterns in Multiple Sequences of Dynamic Graphs

The analysis of a sequence of dynamic graphs plays an important role in many fields.With data accumulated,analysts are able to investigate evolution patterns in multiple sequences of dynamic graphs,which reveals pro-found principles in such dataset.However,the complex structures,massive graphs,and particular evolution process bring great challenges to understand the evolution patterns.To extract leading evolutionary patterns in large volume of dynamic graphs,a new pipeline is proposed.First,we transform graph sequences into a series of event sequences,based on which frequent graph sequences are extracted.Besides,an interactive visual prototype integrating the new pipeline is developed to support top-down exploration of massive dynamic graphs.Finally,two case studies are performed to demonstrate the usability and effectiveness of our method in real-world applications,including kinship network and co-authorship analysis.