Jiezhong Qiu于2016年在WWW会议上发表了题为 The Lifecycle and Cascade of WeChat Social Messaging Groups的论文。1
wechat是一种social messaging groups，基于手机的app应用，虽然市场占有率还没有whatsapp多，但是增长趋势非常惊人。
long-term and short-term groups are inherently distinct.
“Beside studying the friendship structure, we also discuss the group formation processes, namely by investigating the group membership invitation cascading tree structure. We start with defining the group cascade tree below.”
Wiener index, is useful for disambiguating between shallow, broadcast-like diffusion and the deep branching structures.
Wiener index is defined by the average distance between any two nodes in the cascade tree. For example, the cascade trees in Fig- ure 4(a) and Figure 4(b) have Wiener indexes of 3.99 and 1.83, respectively.
For instance, the big dotted circle in Figure 6 encompasses all the current members within a group.
There are two essential steps behind each invitation:
- 1) a member in a group become active (denoted by blue in Figure 6), and
- 2) the active member selects his/her friends (denoted by red in Figure 6) into the group chat.
Invitation Interval is defined as the time inter- val between any two consecutive invitations from a group member.
Additionally, First Invitation Latency is defined as the interval be- tween the timestamp at which a user joins a group (invited by some existing member) and the timestamp when he/she, for the first time, invites another friend to the same group.
we develop a membership cascade model and demonstrate the effectiveness by achieving AUC of 95.31% in predicting inviter, and an AUC of 98.66% in predicting invitee.
Qiu, J., Li, Y., Tang, J., Lu, Z., Ye, H., & Chen, B., et al. (2016). The Lifecycle and Cascade of WeChat Social Messaging Groups. International Conference on World Wide Web (pp.311-320). International World Wide Web Conferences Steering Committee. ↩