NetworkTrekker

 
 

The purpose of this project is to provide scientists with a tool to visualize and analyze networks  in a three dimensional setting. At the present time most of network visualization tools focus on  two dimensional layouts that have been very useful for networks with a moderate number of nodes  (< 1000), however as the size of a given network increases, two dimensional layouts become less  useful. A second motivation of our work is the representation and simulation of real life networks,  where three dimensional coordinates are relevant and are obtained via experiment. Another motivation for creating these network visualization tools is to provide an environment  where the scientists can navigate these networks, an immersive experience that allows them not  only to discover patterns but to ”visit” these patterns and extract relevant information out of  them. This java based visualization tool uses OpenGL to render 3D objects, nodes and links in  our case, in a computer monitor. In addition to that, we have implemented different visual modes  that output the 3D environment in modes acceptable by polarized 3D glasses and by OIT’s 3D  hardware. At the present time we have applied our 3D framework in two cases: Representation of  fibrin networks in blood clots and social networks.

Introduction and Motivation

Fibrin Network

A fibrin network is a mesh that emerges during blood clot formation, it gives stability to the clot and promotes a more stable structure that consequently leads to faster healing. Fibrin network formation has been studied extensively mainly

by experimental groups. Research groups in North Carolina and Indiana University were able to use co-focal microscopy to obtain z-stack images that put together provide a three dimensional picture of a fibril network. These images are processed by Dr. Chen's group in the CSE department in order to obtain spatial coordinates for nodes and connections (fibrins) in this experimentally observed network. We use our visualization tool to interpret these 3D coordinates and provide an interactive environment for analysis and visualization of these experimental networks. This provides an opportunity to scientists to analyze these networks at a closer distances and with the possibility to manipulate perspectives and views. The following movie shows a representation of a reconstructed fibrin web using our viewer.





















The following movie shows an anaglyph representation of the same network ready to be seen with 3D glasses.






























We also produced output that has been used in the 3D hardware of the Center for Research Computing (see pictures of the event and movie below). This project was developed as a project in the Mathematics Department under the supervision of Dr. Mark Alber and Dr. Chris Sweet.
































Social Networks

We used our network visualization framework to represent social interactions in a phone communication network, where nodes represent persons and links represent communication links (calls). Also link width represent the amount of communication as well as its color. Just as in the fibrin network project, the social scientist can traverse this social constellation and obtain insightful  information about it. We can represent networks using standard algorithms for network layouts like Kamada-Kawai or Fruchterman-Reingold. In addition to this, one of our contributions was to provide a new layout where nodes are constrained to the surface of a sphere. Then we model the links as if they were springs with a coefficient proportional to the amount of communication and nodes with electrostatic repulsion forces. We restrict the nodes to be located to parts in the surface of the sphere to generate continents that congregate interconnected subgraphs. This continental view provides a more organized view of large social networks containing thousands of nodes. The following movie  illustrates the output of the viewer. These images can be viewed in 3D displays to augment the analysis experience and interactivity of the network. This project has been developed for the Interdisciplinary Center for Network Science and Applications (iCENSA) in collaboration with Dr. Zoltan Toroczkai.





Initial Phase of Continental Layout Algorithm






























Midle Phase of Continental Layout Algorithm






























Final Phase of Continental Layout Algorithm