Analysing human networks:Part 1
Dr. Ali Fisher
SOAS - March 2014
•History, •American Studies, •Public Diplomacy,•Counter terrorism, •Public health
Fields:
The open source JihadIntroduction
• Why do we think networks matter?
• What type of questions might we ask?
• What do we need to be able to analyse them?
(http://bit.ly/1jMb39a)
The open source JihadFinding relational data
• Interviews
•Letters and archives
• WWII SIGINT (e.g. from Bletchley
Park)
• Familial ties in Renaissance Florence
• Eurovision voting
• Transmission of disease
• ENRON email trail
• Online communities
Researching human networks : Getting Started with relational data
• A sends letter to B
• A marries B
• A retweets B
We are looking for:
Two or more connections between three or more points
Note: Does direction matter in your data?
These connections can be any type of relational data:
Create an edge list with the aggregated connections
Researching human networks : Getting Started with relational data
How to work with our data:
NodeXL – just enter data or use data import
Alternatives; • Create CSV file in spreadsheet or notepad
• Import into Gephi using CSV upload in datalab.
• Convert to a Pajek file using TXT2Pajek (http://bit.ly/1dTr7kF)
Layout matters:Hierarchies in human networks:
•These are the same connections in different layouts
Hierarchies are a type of network
Understanding a network:
What makes someone important?
• Betweenness• Closeness• Degree• Eigenvector• Pagerank
Network data:
• Size - (number of nodes & edges)• Does the network comprise a number of distinct communities?• Density, what proportion of potential connections exist?
Some options:
(there are many others…)
Resources:
Remember: Search engines are your friend…
https://github.com/ptwobrussell/Mining-the-Social-Web
Code on Github:
Example: Open source relational data
66 important Jihadist accounts on Twitter
http://bit.ly/14kYuuI http://bit.ly/1cFbjDg
Check out the interactive version:View full size infographic:
http://bit.ly/1a8Og2I
Read the article
Tracking Response
Information sharing network over time
http://bit.ly/GZ43H8
Example: Open source relational data (part 2)
The open source JihadStarting point for digital data:
“the potential in the era of big data
comes not from drowning in a sea
of data but navigating the most
useful ways to derive insight
and develop innovative strategies”.
Volume, speed, credibility –> emergent behaviour
Focus on the signs which are meaningful to the humans you are studying?
The open source JihadNext time:
• Your projects and questions
• Bigger data examples
• Analytics and visuals