IDENTIFYING THE BUILDING BLOCKS OF SOCIAL SIMULATION MODELS: A QUALITATIVE ANALYSIS USING OPEN-SOURCE CODES IN NETLOGO

Repeatedly developing models from scratch raises the possibility of code bugs and may be an ineffective use of resources. While reusable building blocks or component libraries are available in multiple simulation software and commonly used in industrial engineering, they are often absent from social simulation models or only support data-centric tasks (e.g., import/export maps). We examined social simulation models for human migration (which emphasize the geographical aspect) and rumor spread (which focuses on information processing) to establish whether building blocks could be created across application domains. Based on 39 NetLogo models collected from two platforms (Github and CoMSES), our quantitative analysis with 11 software metrics confirmed the possibility of using building blocks given the current code complexity and time investment of modelers. Our qualitative thematic analysis found five themes in each application domain, of which three were shared and can be a priority to develop reusable building blocks.