Keywords: Simulation, AgentBasedModelling, MultiModalData
Need: In general, ABMs have a solid methodological foundation. However, the implementation and subsequent documentation is inconsistent. For researchers, documenting the degrees of freedom in ABMs is not a primary focus. As such, bespoke models do not explain or share how:
Pertinent information can be captured by a polyglot library that provides metadata about an ABM at runtime and on completion. The ability to describe, annotate and introspect ABM models at runtime, via an opinionated framework, would go a long way in providing standard outputs to validate research and determine the veracity of the model. As well as, documenting the implementation of a particular ABM.
Current Knowledge/Examples & Possible Techniques/Approaches: Currently, documentation is disseminated through articles, conference proceedings or wikis. The models are shared via GitHub, GitLab and OpenABM/CoMSES. The models are written in many languages. As such, the requisite familiarity with the programming language (or tool e.g. NetLogo) used to implement the model is a significant barrier to entry.
Using standardised description formats such as the ‘Overview, Design concepts and Details’ (ODD) protocol is also promoted in the literature. A possible approach would be to implement a library that can generate metadata (at runtime) and a manifest (after each run) for any model with a framework utilising a standardised description protocol.
See for instance the paper Comprehensive and Comprehensible Data Catalogs
Related Previous Internship Projects: N/A as first iteration of the project
Enables Future Work: We aim to use the output of this research for the ABM engine and SynPathGo.
Outcome/Learning Objectives: A possible outcome of this project is a library framework for documenting and labelling ABMs with information at rest and at runtime to generate metadata and a manifest after running the simulation.
Datasets: The project will require existing simulation models in Python to compare the existing documentation against the metadata output.
Desired skill set: Proficient in:
Familiar with:
Return to list of all available projects.