7911 Computational Models for Social Network Analysis -mining big social networks(Part l: Group and Structure) Jie tang Computer Science, Tsinghua University @WW2017
1 Jie Tang Computer Science, Tsinghua University @WWW’2017 Computational Models for Social Network Analysis —mining big social networks (Part III: Group and Structure)
Roadmap data Heterog U ser Tie Structure eneous Micro Macro tie Dynamic nfluence A User modeling Social tie/link Triad Formation Big&Big Demographics Homopt Communit Social role Social Influence Group Behavior social Social Theories Graph Theories BIG Networks
2 Roadmap User Tie Structure tie Influence - User Modeling - Demographics - Social Role - Social Tie/Link - Homophily - Social Influence - Triad Formation - Community - Group Behavior BIG Networks Social Theories Graph Theories Big&Big Dynamic Heterog eneous data social Micro Macro
Roadmap data Heterog U ser Tie Structure eneous Micro Macro tie Dynamic fluence User modeling Social tie/link Triad Formation Big&Big Demographics Homopt Community Social role Social Influence Group Behavior social Social Theories Graph Theories BIG Networks
3 Roadmap User Tie Structure tie Influence - User Modeling - Demographics - Social Role - Social Tie/Link - Homophily - Social Influence - Triad Formation - Community - Group Behavior BIG Networks Social Theories Graph Theories Big&Big Dynamic Heterog eneous data social Micro Macro
Two Questions How groups are formed in the social networks and what is the lifecycle of the different groups? How different people influence the information diffusion in the network- -structural hole
4 Two Questions • How groups are formed in the social networks and what is the lifecycle of the different groups? • How different people influence the information diffusion in the network—structural hole
Group chat in Wechat Group(5) ncel Group Chat 圈2 A Not Set Lukens Orthwein BGDHJKLNoP参w Luke Group OR Code Steve Group Capacity Christy Invite Sticky on Top Save to Contacts Mohammed (a) We Chat group membership(b) Membership invitation -2. 3 million groups generated every day .>25% messages are generated in group chats
5 Group Chat in WeChat •~2.3 million groups generated everyday •>25% messages are generated in group chats Steve Mohammed Luke Christy Invite Invite
WeChat data Group: groups generated on July 26th, 2015 User: group members users in fringe Inv itation: u, v, C, T) Friendship: u, V, T) Category Type Number Group Total 474.726 Member Grou Min group size M ax group size 500 Fringe User nvitation T0201357 friendi Friendship Total624,529,005
6 WeChat Data Fringe Group Member • Group: groups generated on July 26th, 2015 • User: group members + users in fringe • Invitation: (u, v, C, T) • Friendship: (u, v, T) Invite friend
Group Lifecycle Dichotomy Definition Gro a group is initial sends chat How long would a group chat survive? 10 0.4 0.2 10 00L 51015202530 Group Lifespan(day) Group Lifespan(day) Short-term group vs Long-term group
7 Group Lifecycle Dichotomy Definition: Group Lifespan. Duration from the timestamp at which a group is initialized, to the timestamp at which no group members sends chat messages anymore. Short-term group v.s Long-term group How long would a group chat survive?
Group Lifecycle Dichotomy Case Stud Table 2: Case study by group displayed name. Category Long Short Example Travel Discuss on a short trip Meeting 2 Schedule an official meeting Event Entertain 01459 13 Plan a wedding 13 Dine together Organization 0 Departments of company Cla 12 Course for gre Friend 13 0 Childhood friend Family 16 0 A family of three Short-term group v s Long-term group Event-driven v.S. Relationship-driven
8 Group Lifecycle Dichotomy – Case Study Short-term group v.s Long-term group Event-driven v.s. Relationship-driven
Group Lifecycle Dichotomy -Structure Dynamics Open Triad 10 losed jad Long-term · Long-tem Short-term · Short-term 10 10 10 #Open triad( setting up) #Closed triad( setting up) (a) Example (b)Open triads (c)Closed triads Long-term Group: Strong dynamics in terms of underlying friendship structure Short-term Group: Less likely to develop friendship over time
9 Group Lifecycle Dichotomy – Structure Dynamics Open Triad Closed Triad • Long-term Group: Strong dynamics in terms of underlying friendship structure. • Short-term Group: Less likely to develop friendship over time
Group Lifecycle Dichotomy -Group Cascade Tree Definition: Group Cascade Tree. a directed graph where each group member is a node, and a directed edge from u to v is constructed if u(inviter) successfully invites v(invitee) to the group Example of long-term grou WeChat Group Group Cascade Tree →自中 5。 Example of short-term gro
10 Group Lifecycle Dichotomy – Group Cascade Tree Definition: Group Cascade Tree. A directed graph where each group member is a node, and a directed edge from u to v is constructed if u (inviter) successfully invites v (invitee) to the group. Invite Group Cascade Tree Invite WeChat Group Example of long-term groups Example of short-term groups
