Storyline visualizations, which are useful in many applications, aim to illustrate the dynamic relationships between entities in a story. However, the growing complexity and scalability of stories pose great challenges for existing approaches. In this paper, we propose an efficient optimization approach to generating an aesthetically appealing storyline visualization, which effectively handles the hierarchical relationships between entities over time. The approach formulates the storyline layout as a novel hybrid optimization approach that combines discrete and continuous optimization. The discrete method generates an initial layout through the ordering and alignment of entities, and the continuous method optimizes the initial layout to produce the optimal one. The efficient approach makes real-time interactions (e.g., bundling and straightening) possible, thus enabling users to better understand and track how the story evolves.
This work was presented in IEEE InfoVis 2013.
Project page:
http://www.ycwu.org/projects/infovis13.html
23. Discrete and Continuous optimization
Discrete optimization
Continuous optimization
– Edge crossings
– Wiggle distance
– Number of wiggles
– White space
27. Compaction
Quadratic programming
ne nt 1
ne
i 1
i 1 j 1
nt
Minimize ( yi , j yi , j 1 )2 yi2, j
j
Subject to
yi1 , j yi2 , j ,
if
Si1 , j Si2 , j ;
Line order
yi , j yi , j 1 ,
if
Si , j Si , j 1 ;
Line alignment
yi , j yi 1, j din ,
if
SID( Si , j ) SID( Si 1, j );
Line adjacency
yi , j yi 1, j d out , if
SID( Si , j ) SID( Si 1, j ).
Line separate
41. US 2012 Presidential Election
– 2012 US presidential election Twitter Data
• 89,174,308 tweets from May 01, 2012 to November 20, 2012
• 900 users: politicians (334), media (288), and grassroots (276 )
• Two-level location hierarchy
– Five hot topics: Welfare, Defense, Economy, Election, and Horse race
– 2,344 hot hashtags
• Session List
ID
Hashtag
Start
End
Members
0
Hashtag1 140
167
Opinion leader A, Opinion leader B
1
Hashtag2 145
180
Opinion leader C, Opinion leader D
42. Overall Patterns (1/2)
Five significant events on Election
– First debate, VP debate, second debate, and third debate
Grassroots
Media
Political Figures
Defense
Election
First debate
VP debate
Second debate
Third debate
Voting
Economy
Welfare
Horse Race
Timeline
43. Overall Patterns (2/2)
Three user groups focused mainly on Election
– Grassroots also focused on Economy and switched frequently
– Political figures were more focused
– Media occasionally switched
Grassroots
Media
Political Figures
Defense
Election
Economy
Welfare
Horse Race
Timeline
44. Significant Transition
Transition from Election to Economy
Grassroots
Media
Political Figures
Defense
Election
First debate
VP debate
Second debate
Third debate
Voting
Economy
Welfare
Horse Race
Timeline
Sensata
tlot
teaparty
gop
think Romney is tough on china? ask the
workers of #sensata about that as they
train their Chinese replacements
45. Significant Transition
Transition from Election to Economy
Grassroots
Media
Political Figures
Defense
Election
First debate
VP debate
Second debate
Third debate
Voting
Economy
Welfare
Horse Race
Timeline
Issue-attention cycle
sandy
fema
47. Conclusion
A Storyline visualization system
– An efficient hybrid optimization approach
– A hierarchy-aware storyline layout
– A method for interactively and progressively rendering
Future improvements
– Flashback narrative
48. Acknowledgements
Prof. Jonathan J.H. Zhu @ CityU, Hong Kong
Prof. Tai-Quan Peng @ NTU, Singapore
Prof. Kwan-Liu Ma and Yuzuru Tanahashi @ UC Davis