Handling big data on agent-based modeling of Online Social Networks with MapReduce

Abstract

There is an increasing interest on using Online Social Networks (OSNs) in a wide range of applications. Two interesting problems that have received a lot of attention in OSNs is how to provide effective ways to understand and predict how users behave, and how to build accurate models for specific domains (e.g., marketing campaigns). In this context, stochastic multi-agent based simulation can be employed to reproduce the behavior observed in OSNs. Nevertheless, the first step to build an accurate behavior model is to create an agent-based system. Hence, a modeler needs not only to be effective, but also to scale up given the huge volume of streaming graph data. To tackle the above challenges, this paper proposes a MapReduce-based method to build a modeler to handle big data. We demonstrate in our experiments how efficient and effective our proposal is using the Obama's Twitter network on the 2012 U.S. presidential election.