TY - GEN
T1 - A scalable framework for modeling competitive diffusion in social networks
AU - Broecheler, Matthias
AU - Shakarian, Paulo
AU - Subrahmanian, V. S.
PY - 2010/11/29
Y1 - 2010/11/29
N2 - Multiple phenomena often diffuse through a social network, sometimes in competition with one another. Product adoption and political elections are two examples where network diffusion is inherently competitive in nature. For example, individuals may choose to only select one product from a set of competing products (i.e. most people will need only one cell-phone provider) or can only vote for one person in a slate of political candidate (in most electoral systems). We introduce the weighted generalized annotated program (wGAP) framework for expressing competitive diffusion models. Applications are interested in the eventual results from multiple competing diffusion models (e.g. what is the likely number of sales of a given product, or how many people will support a particular candidate). We define the "most probable interpretation" (MPI) problem which technically formalizes this need. We develop algorithms to efficiently solve MPI and show experimentally that our algorithms work on graphs with millions of vertices.
AB - Multiple phenomena often diffuse through a social network, sometimes in competition with one another. Product adoption and political elections are two examples where network diffusion is inherently competitive in nature. For example, individuals may choose to only select one product from a set of competing products (i.e. most people will need only one cell-phone provider) or can only vote for one person in a slate of political candidate (in most electoral systems). We introduce the weighted generalized annotated program (wGAP) framework for expressing competitive diffusion models. Applications are interested in the eventual results from multiple competing diffusion models (e.g. what is the likely number of sales of a given product, or how many people will support a particular candidate). We define the "most probable interpretation" (MPI) problem which technically formalizes this need. We develop algorithms to efficiently solve MPI and show experimentally that our algorithms work on graphs with millions of vertices.
UR - http://www.scopus.com/inward/record.url?scp=78649253525&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=78649253525&partnerID=8YFLogxK
U2 - 10.1109/SocialCom.2010.49
DO - 10.1109/SocialCom.2010.49
M3 - Conference contribution
AN - SCOPUS:78649253525
SN - 9780769542119
T3 - Proceedings - SocialCom 2010: 2nd IEEE International Conference on Social Computing, PASSAT 2010: 2nd IEEE International Conference on Privacy, Security, Risk and Trust
SP - 295
EP - 302
BT - Proceedings - SocialCom 2010
T2 - 2nd IEEE International Conference on Social Computing, SocialCom 2010, 2nd IEEE International Conference on Privacy, Security, Risk and Trust, PASSAT 2010
Y2 - 20 August 2010 through 22 August 2010
ER -