Agent-based SImulation, MOdeling, and Visualization of processes
Suki van Beusekom
(keywords: agent-based simulation, process analysis, knowledge discovery)
ASIMOV infers knowledge about the details of a process (e.g., consultation) occuring in a context (e.g., hospital) with the sole generic information about the process itself and measurement about a real-life process of that kind. Such knowledge can then be used to calculate KPI's, visualize analytics, make predictions, validate models, explain processes, etc.
For instance, the details about the sequence of activities (e.g., usage of resources, movement of people, probability of certain events occuring) can be inferred out of a sensor network measuring occupancy, temperature of a specific context and some generic information (e.g., kind of activities etc.) about the generic process.
Given a business process described as a flowchart and some measurements about the process, ASIMOV can infer what sequence of events have led to the measurements, thus inferring and abstracting knowledge about the process itself.
Estimated TRL/Maturity: TRL 4 (i.e., technology validated in lab), medium maturity (tests and documentation still needed).
Past projects: ARUM, ADAPT4EE
Unique selling points: open source, highly flexible, possibility to be applied to any process (potentially huge horizon of applications), validation vs. real or synthetic data.
Competitors: unknown at the moment.
Key references: ADAPT4EE documentation, van Beusekom et al PAAMS 2015 (in preparation)
Demos available: ADAPT4EE use cases demo
Possible developments: further documentation, validation of current results, further machine learning algorithms to infer knowledge, more integration with Eve and vis.js
Ideas for application: energy and mobility projects in Smart Cities (this can be used as a tool for massive agent-based simulation), factory of the future calls (big applications in manufacturing), or anything that has processes and data/measurements in it.