How To Find Key Opinion Leaders: The #IstandwithCEU Campaign

Originally published on CEU DNDS blog in 2017, co-authored by Koman Zsombor.

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).

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. 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).

We have found that even though naive intuition would suggest that highly followed users (up to a several million followers) are the most important in achieving wide social reach, there is a number of less popular users (with around 10k followers) among the key opinion leaders. This means that besides high general recognition, a number of quality tweets and a reasonable amount of influential followers can largely enhance social reach by inducing waves of retweets among influential users.

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.