The impact of common neighbor algorithm on individual friend choices and online social networks

Online social platforms have become increasingly more popular for people to make new friends, by relying on the friend recommendation algorithms implemented on social networks. Nevertheless, how these algorithms impact the choice of friends of individual users and the structure of overall social networks remains unknown. In this paper, we introduce a model in which a group of users interact and make friends as recommended by the common neighbor algorithm, which is one of the most commonly used friend recommendation algorithms, to study the impact of the algorithms on the choice of friends of users in social networks. Based on our results, we found that the algorithm is mostly effective in identifying good matches, but users may group themselves into sub-optimal clusters when they over-rely on the algorithms. These results demonstrate the pros and cons of the increasingly more popular common neighbor algorithm applied in social networks. The model is then examined with the attribute similarity matrix obtained from two real datasets, and the results are consistent with our earlier findings. We also investigate the impacts of user reputation on the common neighbor algorithm and found that users with high reputation may become network hubs connected with a majority of users on the platform. Despite the simplicity of our developed model, our results provide interesting insight into the impact of common neighbor algorithm on friend choices of users and the global characteristics of social networks.