Authors:
Mert Ozkaya
1
;
Mehmet Kose
2
and
Egehan Asal
3
Affiliations:
1
Yeditepe University, Department of Computer Engineering, Istanbul, Turkey
;
2
Independent Researcher, Istanbul, Turkey
;
3
DFDS, Istanbul, Turkey
Keyword(s):
Architecture Description Languages, Event-Driven Architectures, Decision Support, Simulation, Modelica.
Abstract:
Decision support systems (DSSs) have been existing for automating the decision making processes and reaching the optimum decision(s) using a data set in the quickest way. Despite the importance of DSSs, no any architecture description language (ADL) have been proposed for the high-level specifications and analysis of DSS architectures. So, in this paper, we propose a new ADL called DecSup which enables for the graphical specifications of DSS architectures in terms of the problem, diagnosis, and action components that interact with each other in an event-based manner. Problem components represent the domain data sets whose initiali-sation/change trigger an event for the diagnosis component. Diagnosis components include pattern predicates for making diagnosis using the events occurring. Whenever a diagnosis is made, another event is emitted for the action components to take any necessary actions. DecSup is supported with a prototype toolset for specifying the architecture models and tr
ansforming models in the Modelica simulation language. The transformed Modelica code can be used to simulate the DSS architecture models and test the architectural decisions via some scenarios. We evaluated DecSup using a case-study based on the contagious respiratory illnesses (i.e., cold, flu, and Covid-19).
(More)