The thought leader algorithm

In each of our news communities we curate a list of top stories by first algorithmically identifying subject matter experts who are respected amongst their peers.

How the rankings are calculated

In simple terms, thought leaders are calculated based on the following:

  1. How many thought leaders follow and retweet you
  2. How much apparent interest do you have in the topic (determined by who you follow and what you tweet about)

Note. Thought leaders are currently only for people with a Twitter account.

The detailed version

Our algorithm starts out with a set of seed users (either manually or algorithmically defined) and assigns them equal weight. Looking at Twitter follow and retweet data we calculate the most respected people amongst the seed users. From here we adjust the weights and iterate the algorithm, getting a better and better idea of who the thought leaders are with each pass. The algorithm is similar to Google's page rank algorithm -- the difference is we're looking at links between people instead of web pages.

Seed users are defined depending on which type of community it is:

  • Seeding of Topic Communities: These communities center on a topic, and in this case we're able to generate the list of thought leaders entirely from a set of seed keywords. The algorithm may have other systemic biases -- but the only community specific bias being introduced by Thinklab is selection of the seed keywords.

  • Seeding of Cause Communities: These communities can be more political in nature and thus we may have "seed users" defined to help the algorithm identify the people that are generally supporting the cause. This introduces a certain amount of bias -- but it should be noted that on an individual level, being defined as a seed user only affects a persons ranking slightly.