Chair of Digital Marketing

New publication on user targeting in Journal of Interactive Marketing

Simon Stolz and Prof. Dr. Christian Schlereth show in their new article “Predicting Tie Strength with Ego Network Structures” how network structures can be used to identify the closest friends. The method has a high potential for user targeting and has important implications for data privacy.

In social networks, like Facebook, LinkedIn, and XING, we usually have a high number of contacts – close friends but also many acquaintances. The authors Simon Stolz and Christian Schlereth show in the publication “Predicting Tie Strength with Ego Network Structures” that it is possible to identify the rare closest contacts of a person by using ego network structures. In the publication, 18,541 individual connections of Facebook are classified in “closest friends” and “no closest friends” by using a Machine Learning approach. While the previous literature has mostly used predictors, such as the number of common friends, the article shows that especially bridging positions within ego networks are an important indicator for a strong tie (i.e., closest friends). The bridging position within ego networks often bridge between different social circles (e.g., family and university). While the members within a social circle already know each other (e.g., all family members know each other) the bridging position is an indicator that the connection may have been established through an introduction of the focal user. In combination with further predictors, the rare closest friends can be identified with a precision of 45%. The article appeared in Journal of Interactive Marketing, Volume 54, 2021 and is openly accessible until 21st January via this open access link.

Most importantly, with this research, we want to share a rich data set: The data contains not only the ego networks but also who are the closest friends for each ego. The anonymized data set has been published on Mendeley Data for further exploration. The data also constitutes an interesting example for teaching social network analysis. Also, a presentation of the paper is accessible via slideshare.