D. Di Caprio, F. J. Santos Arteaga, M. Tavana
We use a self-organizing map neural network algorithm to analyze the formation of social networks determined by the preferences of their users, who are endowed with incomplete information regarding the characteristics of other users from who they receive friendship requests. The acceptance or rejection decision is determined by the limited information available when receiving the requests, the expectations regarding the remaining characteristics of the requesters and the resulting improvement in network capacity derived from accepting the requests. We illustrate how the similarity in preferences among users leads to more cohesive networks within the incomplete information scenario analyzed. Moreover, the emergence of disutility costs derived from a suboptimal decision when accepting a request increments the separation between clusters. In this regard, the inclusion of users endowed with average preferences serves as a cohesion mechanism that reduces the distance between clusters.
Keywords: Social Media, Expected Utility, Preference Similarity, Self-Organizing Map, Neural Networks.
Scheduled
TE1 Computational Management Methods
May 31, 2016 4:45 PM
Salón de actos
