Abstract

Social Network Sites (SNSs) like Instagram have proven to be effective platforms for Word-Of-Mouth marketing (WOM), and simulations of Agent-Based Models (ABMs) can predict users' behavior in these networks. Using ABMs allows researchers to understand real-world scenarios by modeling and abstracting them, and to extract and identify the essential behaviors to study. To develop an ABM, a programmer can follow two approaches. First, developing a program from scratch that runs a specific ABM simulation. Second, using known frameworks like Repast to speed up implementation. The issue with the first approach is that advanced programming knowledge in ABMs is required, and with the second one, apart from ABM knowledge, a programmer needs a deep understanding of these frameworks. This paper proposes the architecture of a Flexible Agent Simulator for Open WOM (FASOW), which uses a design strategy inspired by the concepts associated with Reflective Tower to provide advanced modularity features and a less steep difficulty curve for developers. This architecture allows developers to decompose a complex model into layers of varying complexity, gradually increasing the required knowledge. To demonstrate the effectiveness of FASOW, we replicated two published ABMs. Additionally, we provide an online, functional proof of concept to showcase the architecture's practical applicability.