|
|
 |
| |
|
|
|
Scalable Visualization of Semantic Nets using Power-Law Graphs |
|
|
|
PP: 355-367 |
|
|
|
Author(s) |
|
|
|
Ajaz Hussain,
Khalid Latif,
Aimal Tariq Rextin,
Amir Hayat,
Masoon Alam,
|
|
|
|
Abstract |
|
|
| Scalability and performance implications of semantic net visualization techniques are open research challenges. This paper
focuses on developing a visualization technique that mitigates these challenges.We present a novel approach that exploits the underlying
concept of power-law degree distribution as many realistic semantic nets seems to possess a power law degree distribution and present
a small world phenomenon. The core concept is to partition the node set of a graph into power and non-power nodes and to apply
a modified force-directed method that emphasizes the power nodes which results in establishing local neighborhood clusters among
power nodes. We also made refinements in conventional force-directed method by tuning the temperature cooling mechanism in order
to resolve ‘local-minima’ problem. To avoid cluttered view, we applied semantic filtration on nodes, ensuring zero loss of semantics.
Results show that our technique handles very large scale semantic nets with a substantial performance improvement while producing
aesthetically pleasant layouts. A visualization tool, NavigOWL, is developed by using this technique which has been ported as a plug-in
for Protege, a famous ontology editor. |
|
|
|
|
|
|
|
