Empirical Analysis of Relationship-based User Reposting Behavior on Microblog Network

Abstract

Microblogs have become a significant online social service for information propagation. Compared with other SNS, a microblog makes use of a flexible unidirectional subscription structure to encourage users to get information from others. In this paper, the authors performed relationship-based large-scale statistics and analyzed long-term reposting behavior. The total number of reposts from a certain followee with a bidirectional or unidirectional relationship is power-law, as well as a reposting interval time. Statistics suggest the curve of bidirectional friends decays slower with a smaller slope, which implies the interactions between bidirectional friends are more stable and intensive. Moreover, reposts from bidirectional friends take approximate 55 percentages, although bidirectional relationships comprise only 33 percent of relationships. Furthermore, the authors investigated the impact of following relationships and found that users focus more on social influence and activity level for unidirectional friends; however, for bidirectional friends, interactions and closeness are more crucial.