Network analysis with Gephi

Johan Jarlbrink

Slides of the presentation.

Gephi is an easy-to-use software for network analysis and visualization. The tool can handle large sets of network data of many different kinds: social connections and communication patterns, bibliographic and bibliometric data, word co-occurrences in texts, financial flows and political ties – everything with nodes and edges. This is not a tool for data extraction, you need other tools to collect and curate the data in a spreadsheet format that can be processed by Gephi. But once you have the data you can use it to explore connections and patterns, do statistical analysis, and identify clusters and central components. For smaller and manually compiled data sets the tool can be used to create network graphs.

During the workshop we will go through the basic steps of network analysis in Gephi: data curation and import from Excel, statistical features, filters and visualizations. We will work with two data sets to illustrate different kinds of networks: one with co-occurring words, one with people and their movements.

Download Gephi for free from https://gephi.org/

Download the test datasets here.

 

GEPHI INSTALLATION INSTRUCTIONS:

For Gephi to be able to communicate with Java:
Open the ”gephi.conf”-file in WordPad – search for the file name or find it in the etc-folder
Find where Java is located on your computer and copy the address – on a PC it might look like this:
C:/Program Files (x86)/Java/jre1.8.0_161
(On a PC it is safest to change \ into /)
Go to the file you just opened in WordPad and paste the address to Java after ”jdkhome=”. Add quotation marks around the address and delete the #-sign. On a PC it should look similar to this:
jdkhome=”C:/Program Files (x86)/Java/jre1.8.0_161″
And – when you have the file in WordPad open already – allow Gephi to use more memory. Look for “Xmx0909m” – write Xmx1024m instead. Close and save the file and restart your computer. And hopefully everything will run smooth.

Jarlbrink network

Mainokset