Twitter and social network analysis


It's been a very long time since the historic first Tweet "just setting up my twttr", sent by founder Jack Dorsey, in 2006. In more than 10 years, Twitter has not become the most popular social media platform, yet it still remains an interesting and unique platform to analyze for three main reasons:

  • Organized by interests: Differently from Facebook, where most people follow people they’ve met, Twitter is organized around interest graphs. Connections exist between people and interests as well as between interests and interests.
  • Fully public: Your Tweets are public by default; anyone with an internet connection can view and interact instantly with your tweets.
  • Rich hashtags: With the 140-characters limitation, every character counts, so users carefully select the hashtags they use in their tweets’ text.

Social network analysis (SNA) is an advanced form of analytics that is specifically focused on identifying and forecasting connections, relationships, and influence among individuals and groups. It mines transactions, interactions, and other behavioral information that may be sourced from social media, which may have been previously limited to CRM, billing, and other internal systems.

Why (online) Social Network Analysis

Social network analysis is a useful means of mapping the shape of virtual crowds. For example, visualizing social media conversations with a graph of relations between actors or between contents could help to answer some questions, such as:

  • Who talks with whom?
  • Who is the focal point of an online community?
  • Which communities are involved in particular conversations?
  • What are the more interesting topics?

In 1934, Jacob Levy Moreno, an Austrian-American psychiatrist, was the first scientist to use social network analysis to visually depict relationships between a specific group. In his famous book, Who Shall Survive?, he introduced the “sociogram” to uncover the underlying relationships between people.

Image: One of Moreno’s first sociograms.

Nowadays, large volumes of datasets and tools are available to gather information from social media and create visual depictions.

Relationships can be explicitly stated or be deduced. For example, in social networking platforms, individuals can explicitly declare that they are "friends" (for example, on Facebook) or "linked" (like on Linkedin), join a discussion group (Linkedin or facebook), follow a user (Twitter, Facebook), and more.

Implicit connections can also be identified by a person's activities, for example by analyzing repeated interactions between two or more actors. If we are talking about social networks, this can be seen in sharing actions, liking, tagging, or simply commenting and interacting with content created by a specific user or user groups.

Image: Twitter user relations network.

For example, in the picture above, is represented the network of Twitter user relations around a brand (in the case 5.000 Twitter mentions). Each circle (node) represent a Twitter user name and its diameter I proportional to the number of retweet and mentions received. Two nodes are connected if the users have interacted in the time range analyzed. Different colours represents different communities where interactions are mode dense.

Image: Twitter hashtag copresence network.

Implicit connections can also result from similarity. For example, users using similar tags or tweets around same topics can be considered related, or linked. In the picture above are represented a network of hashtag where two hashtags (nodes) are connected if they are present in the same tweet. Colours represents thematic clusters where same concepts are associated, related, in the words of the people that have written the posts.

So, how can you use social network analysis in your reporting?

In an age of social media, social network analysis is increasingly being used across the communication discipline and is becoming a promising technique for the digital journalist's toolkit.

SNA allows journalists to unveil relationships between individuals and organisations, identify key players and communities of interests by using information on how people and organisations are connected with each other.

To help get you started, I’ve developed an online self-paced course, called Introduction to Social Network Analysis, which teaches the basics of online conversation analysis, with particular focus on Twitter and Facebook. The aim is to provide an overview of the SNA methodology and introduce some entry-level tools for collecting, analyzing and displaying online social relationships through simulations and case studies.