SOCIAL NETWORK INTERACTION QUANTIFICATION AND RELATIONSHIP TREND ANALYSIS WITH MULTI-AGENT SYSTEMS

摘要

Multi-agent systems (MAS) have been used to simulate interpersonal relationships and communication in a variety of studies. Using MAS, we propose the NCRAE (Neighborhood Cumulative Reward Average Evaluation) method to model interactions between individuals in a social network and interpret changes in their relationships over time. By quantifying the effects of call and SMS interactions in a given dataset among our simulation environment, we identify influences that may cause shifts in relationship closeness. Our results show a successful differentiation of close and distant relationships through their respective communication patterns and characteristics. We find that close relationships maintain a high level of interaction frequency and duration with no significant increase in closeness, whereas more distant relationships engage in little to no communication but engage in proximity interaction that defines its changes in relationship closeness.