This dataset contains a social network of Last.fm users that are part of the LFM-1b dataset [1] and has been created using the Last.fm API [2]. More specifically, we have adopted the breadth-first-search sampling strategy with LFM-1b users as seed nodes and crawled friends up to two hops away (friends and friends of friends). We have kept only those edges where both nodes connected by an edge are part of the original LFM-1b dataset. These edges are stored in the 'LFM-1b_social_ties.txt' file which contains 78,989 edges between 11,792 unique users. Additionally, we have stored demographic and playcount information for each of these users in the 'LFM-1b_users.txt' file. This dataset has been used as input for the purposes of studying homophily and link prediction with user preferences for mainstream, novelty, and diversity in music [3] and to reproduce the experiments from that study, we have also provided an open-source GitHub repository [4]. [1] http://www.cp.jku.at/datasets/LFM-1b/ [2] https://www.last.fm/api [3] Duricic, T., Kowald, D., Schedl, M., & Lex, E. (2021, November). My friends also prefer diverse music: homophily and link prediction with user preferences for mainstream, novelty, and diversity in music. In Proceedings of the 2021 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining (pp. 447-454). [4] https://github.com/tduricic/homophily-lastfm