Solving Influence Problems on the DeGroot Model with a Probabilistic Model Checking Tool

Abstract

DeGroot learning is a model of opinion diffusion and formation in a social network of individuals. We examine the behavior of the DeGroot learning model when external strategic players that aim to bias the final consensus of the social network, are introduced to the model. More precisely, we consider the case of a single decision maker and the case of two competing external players, and a fixed number of possible influence actions on each individual. When studying the influence problems, we focus on the stochastic processes underlying the solution of DeGroot problems. In case of one decision maker, the analysis of the DeGroot model leads to the formation of a Markov Decision Process (MDP) and in the case of two external competing players the model is reduced to a Stochastic Game (SG). Since such models are heavily used in probabilistic model checking we apply tools of the field to solve them. Preliminary experimental results confirm the viability of our approach, which relies on the common mathematical foundations of the DeGroot problems and probabilistic model checking.

Citation

@inproceedings{gyftopoulos2016solving,
title={Solving Influence Problems on the DeGroot Model with a Probabilistic Model Checking Tool},
author={Gyftopoulos, Sotirios and Efraimidis, Pavlos S and Katsaros, Panagiotis},
booktitle={Proceedings of the 20th Pan-Hellenic Conference on Informatics},
pages={31},
year={2016},
organization={ACM}
}

BibTeX

@inproceedings{gyftopoulos2016solving,
  title={Solving Influence Problems on the DeGroot Model with a Probabilistic Model Checking Tool},
  author={Gyftopoulos, Sotirios and Efraimidis, Pavlos S and Katsaros, Panagiotis},
  booktitle={Proceedings of the 20th Pan-Hellenic Conference on Informatics},
  pages={31},
  year={2016},
  organization={ACM}
}