Communications of the ACM
Development Use Cases for Semantics-Driven Modeling Languages
Credit: Kentoh
Developing software and software-intensive systems always requires the development of models. Some are formulated explicitly in some kind of modeling language. Others exist only as mental models and are finally represented implicitly by programs. In fact, application software realizes specific models that are available in the application domain and additionally includes specific technical implementation concepts in terms of hardware, operating systems, and further elements of the required software stack. In the end, application software appears to be a complex, entangled mixture of application models and implementation technology. Like programming languages, Modeling languages are defined by their syntax, which describes the form of the language constructs—textual or graphical—and by their semantics, which specifies their meaning. We claim that the choice of the underlying semantic theories and definition techniques must closely follow the intended use cases for the modeling language. The choice of the syntax should be guided by the semantic domain and its underlying theories and not the other way around.
Key Insights
Modeling languages, such as UML, SysML, and various domain-specific modeling languages, have been suggested to support the specification and construction of systems for specific domains. This leads to the following key questions:
- How should we model real-world entities of systems and how can we relate these entities to their models precisely?
- How should we link up the syntactic representations of models in a modeling language with their semantics?
- How can we represent, specify, and use semantics?
No entries found
Log in to Read the Full Article