Gephi: 2D network visualization
By Dave Song
Visualizing Community Networks in VR Tutorial
Often times, it is important to ask "the right question" regarding building and researching VR software. My main focus for the project 1 was on evaluating if the 3d representation of the network cluster graphs can be better represented in VR space. This was done by comparing 2D visualization and VR force directed graphs in VR space. Through this project, my goal was to create my own VR Force Directed Graph generator as well as learning about current state of 2D network visualization technologies. Additionally, by comparing those two technologies/software, I wanted to evaluate if the VR visualization of the community networks will be a valid mode of visualization.
To do so, Gephi was used to create 2D visualization of community networks.
For more information on Community/Cluster graphs as well as some of the algorithms we can use, please take a look at Getting Started with Community Detection in Graphs and Networks
What is Gephi?
Gephi is a powerful network visualization software that enables users to analyze and visualize complex networks. Developed in Java and released under the GNU General Public License, Gephi provides a user-friendly platform for exploring, manipulating, and presenting data in a network format. With Gephi, users can import network data from various sources, apply various layout algorithms to visualize network structures, and customize the appearance of nodes and edges to communicate their findings effectively. Gephi is widely used by researchers, analysts, and data scientists to gain insights into complex systems, such as social networks, biological networks, and transportation systems.
Generating a Community Network Using Gephi
One of the main functionalities of Gephi is generating community/cluster network graphs using various built-in algorithms. This wiki will how to start Gephi and make your first community network graph .
Download Dataset used in this tutorial : please download dataset 1
Pictures are attatched blow
Open a new project
Navigate to Data Laboratory Menu
Click Import Spreadsheet
Find Downloaded File and open DataSet1.
When you get to Import Report, make sure to click Append to existing workspace option.
Once Nodes1.csv has been imported, click on Edges tab and repeat the same process. This time, we are importing Edges1.csv file.
Now we have 14,115 edges and 1000 nodes!
Once we import the nodes and edges to our data table, you will see something similar to the graph below in the Overview section.
From here, follow the following instructions
In the ranking panel on the left side, select Node tab and then select Degree Option. Then enter min size of 10 and max size of 100. Once you enter the values like below, click Apply
2. Now, we have nodes with different size depending on the connectedness of the node to others in the dataset. From here, we are going to apply Fruchterman Reingold as well as Force Atlas 2 layout algorithms to generate a spatialized version of the graph. Previously, all nodes were generated randomly.
Apply Fruchterman Reingold algorithm first. Then, press run and wait until the graph stabilizes. Then stop the algorithm and apply ForceAtlas2. Repeat the stabilization process.
Once the graph has been stabilized, you will have a cluster network graph generated with Gephi! When you run Fruchterman Reingold, your graph will look similar to the graph below.
Applying Force Atlas2 will result in the graph somewhat similar to the one displayed below.
For more in-depth instructions and information on Gephi, here are some useful sources(including the source used to create this graph as well!):
Gephi Youtube Tutorial for working with Twitter mention networks