• Milán Janosov

How To Find Key Opinion Leaders: The #IstandwithCEU Campaign

Updated: Apr 30, 2021

Following our previous analysis, we studied the prevalence of the #IstandwithCEU campaign on Twitter. This includes the examination of 54k tweets made by 15k users over two months with a total social reach of 180M. This project was carried out as a final project for the Science of Success class of Prof. Albert-László Barabási.

We reconstructed the retweet network of the users by introducing a directed link between two users if one retweeted the other. We calculated the importance of the nodes using the so-called PageRank (PR) algorithm proposed by the founders of Google. This method quantifies the level of influence of each user in the given sample. The underlying concept is similar to a voting system: if a user retweets someone then the retweeted user gets endorsed, and the level of endorsement is proportional to the retweeter’s influence (e.g. number of followers).

Fig. 1. The retweet network of the 1000 most influential users active in the campaign. The nodes represent the users, their size is proportional to the number of their followers, and their color encodes PR centrality scaling from white (low importance) to red (high importance).

The analysis shows that the PR score is correlated with the follower count, meaning that the previous success and appreciation is logically related to the current influence and acquired attention. This is supported by the network visualization as well. Coloring tells us that there are users who have many followers but are not so influential (Fig. 1.: large nodes in color white; Fig. 2.: below the dashed line) and those who do not have so many followers but have a great influence (Fig. 1.: small nodes with color red; Fig. 2.: dots above the dashed line).

Fig. 2. The PageRank score versus the follower count of each user (represented by a dot) who tweeted in the #IstandwithCEU topic. The dashed line represents an analytical formula (power law), while the grey dots scattered around it are the binned trend of the data.

Original source: CEU DNDS website

Original date: June 19, 2017

Collaborator: Zsombor Koman

10 views0 comments

Recent Posts

See All