For many problem domains domain-specific languages (DSLs) offer users more appropriate notations and abstractions in which to model systems when compared with general purpose programming languages. These benefits can often be amplified if a visual notation is used instead of textual notations. In many problem domains visual notations are preferred by practitioners as they often are the most intuitive representation of a problem. However, the lack of supporting infrastructure for constructing, implementing, and maintaining visual languages in general and domain-specific visual languages (DSVLs) in particular has been an impediment to gaining wider acceptance. This paper describes techniques used in the Moses tool-suite for defining the syntax and semantics of DSVLs, which are very general, yet are built on a few very simple concepts and are therefore easy to apply.

models An important part of the design of complex systems is the evaluation of the large number of potential alternative designs. Due to the number and complexity of design parameters, this design space is potentially huge and very complex. Automating part of the design exploration task can be an invaluable help in finding the optimal or near optimal settings of design parameters. The choice of the most appropriate exploration strategy depends on the nature of the parameters, such as their role in the model, the dimensionality and structure of the design space including the number and location of local optima, etc. This paper advocates the use of higher-order modeling techniques to express exploration strategies. This allows users to formulate them in the same set of languages used to model the original system. Hence the set of design space exploration tools can be extended and parameterized as easily as the model itself. In this paper a higher-order modeling langage is presented. As an example a number of simple exploration tools are modeled and applied to a small optimization problem. 1

Models built using visual forms which are representations of the domain concepts are easier to be understood and expressed by the people who work in this domain. Many projects produce modeling environments that offer only the domain concepts to the user but with a single graphic view that the user has to be satisfied with. In this paper, we present our framework for producing domain-specific modeling tools. This framework is independent of the graphic view. So, it can associate for the same domain several graphic views. Without focusing on the modeling graphical interface itself, this article presents a solution to support the generation of such interfaces. In addition, this framework improves the potential re-usability of view and domain descriptions. 1