Social Network Analysis

SNA

What is SNA ?

Social Network Analysis (SNA) boils to one concept: our relationships, taken together, define who we are and how we act. Our personality, education, background, race, ethnicity all interact with our pattern of relationships that we are embed in and leave marks on it. Observing and studying these patterns we can answer many questions about our sociality.

SNA is the study of human relationships by means of graph theory.

SNA formalizes human interactions (and/or relations) and allows to infer insights, in the same way that statistical tools allow us to project the future given the past observations.

Why is SNA timely ?

Internet scaled up the number of human interactions significantly, we are no longer cooperating only with the group of people living physically in same area. Also it has deepened the reach, is equally as easy to talk to a close relative as it is to a distant acquaintance.

A piece of information can now spread faster than ever across a network of people and is only getting faster. And the potential of wisdom of the crowds (“Many Are Smarter Than the Few”) is only getting bigger.

Social actions in digital form most often reflect non-digital life, who you network with and what personality/culture you emerged in.

Social actions happening in digital form, make it possible to collect as data for analysis.

SNA “for Fun and Profit”

i.e. SNA Applications

Who in a network is an influencer?

If you find in the network who everybody else listens to, you can then predict some of the future dynamics of the network, including virality.

Advertisers and Marketers can estimate the biggest bang for the marketing buck by routing their message through the most influential people in a Social Network.

Google page rank algorithm represents the html web pages connections as a network and finds what pages are the authorities on a specific topic, these are the ones to show up first in a results of a search.

Who bridges between groups?

The node that bridges between groups holds an important position as the communications bottlenecks and gets filtered by this person. Is a special type of influencer too.

Terrorists cells are known to operate in this way, as part of the information filtering.

Also If this person leaves the network these 2 groups will become separate and the network weaker in consequence.

If you have a commercial product you can infer that with this person leaving, the loss is more significant (than with a non-bridge) and you’ll want to re-enforce the network to not depend on 1 bridge only.

What different groups are there?

Separate groups of users are an indication of different interests/behaviors/cultures. This enables more personalized (and effective) marketing and/or targeted content in the commercial space.

Also people in groups influence each other, and they often move together, imagine the case of joining an app or leaving an app, whole groups will move.

Friend of a Friend

Typical in a commercial social app, to have recommendations like: “John is a friend with 2 of your friends, maybe you want to get in touch with John too”.

Friend of friend tend to have close interests/cultural profile, so there’s similarity inferences to be made here.

Shortest route between A and B

Imagine you want to move a message between A and B, the graph view can calculate the shortest path.

Finding nodes with the shortest distance to all other nodes inside a group (of nodes), means is the most efficient way to spread a message. Like, “Linda is the one talking to everyone in the social circle and organizing the night out’s.”

How does a network really works ?

Knowing influencer, bridges, groups, FoF, densities, distances etc… Allows to understand how information spreads, how fast , who play’s what roles, how people are really using the app, and what for…. At the end of the day it translates into a competitor advantage on how to optimize a network.

Other

The application of SNA varies a lot, if we study a political network we will focus on different aspects compared to a social network and these translate into different uses, but above are typical concepts that illustrate potential.

Naturally many new applications are appearing and some of the bigger gems happen with cross field pollination’s, for example an obvious synergy is with leveraging psychology principles and social networks on the internet.

SNA in the News

  • The Facebook Scandal - or how to predict psychological traits from Facebook likes: https://www.liip.ch/en/blog/the-facebook-scandal-or-how-to-predict-psychological-traits-from-facebook-likes
  • Palantir has secretly been using new orleans to test its predictive policing technology: https://www.theverge.com/2018/2/27/17054740/palantir-predictive-policing-tool-new-orleans-nopd

References

  • https://en.wikipedia.org/wiki/Social_network_analysis
  • Book: Social Network Analysis for Startups, Alexander Kouznetsov and Maksim Tsvetovat
  • https://medium.com/@himanshubeniwal/a-network-analysis-of-game-of-thrones-73360505e541
  • https://en.wikipedia.org/wiki/Twitter_Revolution
  • Using Palantir to Address Information Security and Insider Threat in the Enterprise, Video Demo: https://www.youtube.com/watch?v=dGPDLD0tJxI&list=PLCA98B156F7EFD6A0
  • An Introduction to Graph Theory and Network Analysis (with Python) https://www.analyticsvidhya.com/blog/2018/04/introduction-to-graph-theory-network-analysis-python-codes/

Updated: