Hierarchical social network analysis using multi-agent systems: A school system case

The quality of K-12 education has been a major concern in the nation for years. School systems, just like many other social networks, appear to have a hierarchical structure. Understanding this structure could be the key to better evaluating student performance and improving school quality. Many studies have been focusing on detecting hierarchical structure by using hierarchical clustering algorithms. We design an interaction-based similarity measure to accomplish hierarchical clustering in order to detect hierarchical structures in social networks (e.g. school district networks). This method uses a multi-agent system, for it is based on agent interactions. With the network structure detected, we also build a model, which is based on the MAXQ algorithm, to decompose the funding policy task into subtasks and then evaluate these subtasks by using funding distribution policies from past years and looking for possible relationships between student performances and funding policies. For the experiment, we use real school data from Bexar County's 15 school districts in Texas. The first result shows that our interaction-based method is able to generate meaningful clustering and dendrograms for social networks. Additionally our policy evaluation model is able to evaluate funding policies from the past three years in Bexar County and conclude that increasing funding does not necessarily have a positive impact on student performance and it is generally not the case that the more is spent, the better.