Twitter influencer analysis
Asset experts researched a method to highlight the most visible Twitter users on a given topic, and wrapped the mathematical analysis on a product ready Web application. The method aims to highlight the user visibility on the discussion, therefore it relies on actions that clearly determine connections among users that are also visible to other users. Tweets are collected via the Twitter streaming APIs and, at a later time, they are processed in order to retrieve a list of users sorted by visibility in the current discussion.
Driven by Engagement Actions
The complex network of Twitter users is derived from engagement actions, like the latest retweets and citations among active stakeholders. As result, the method is effective on monitoring discussion dynamics on the short period related to topics or recent events.
Focused on Active Communities
The most relevant active communities on the topic are found by means of algorithms from Graph Theory. Communities are mainly composed by stake holders accounts, like researchers, journalists, politicians and so on. Institutional accounts may appear when they are engaged by other users.
Ordered by Visibility
Twitter users are sorted according to their visibility, which derives from a Finite Markov Chain model. Most visible users are present first, along with information about the users they are most connected with.
The method is designed to operate on data streamings and to produce new results on a daily basis. It continously observes the discussion on the topic of interest and it returns the most relevant tweets exchanged.
A first report: Zika and vaccines
An analysis of the most relevant tweets and accounts about some key words, with a main focus on Zika virus and vaccines, have been performed, and the corresponding report can be found here.