Search

A complete information modeling method must address both the process- and data-perspectives, preferably in an integrated manner (i.e., also specifying which state change each process should achieve exactly). However, most approaches either emphasize only data or only processes. And when an approach handles both data and processes, there is usually no integration of processes and data. A language must have a precise semantics if tools are to perform intelligent operations on models expressed in the language. Moreover, formal semantics helps in detecting errors in and reasoning about specifications. We give a precise, declarative, model-theoretic semantics for a large class of instruction languages that treats processes and data in an integrated manner. The instruction expressions are interpreted against a ‘state space’ (i.e., a set of ‘states’) and we consider the semantics of an instruction as the set of possible state transitions it can achieve. As a result, we can provide the integration of the different scenarios of a use case into one (textual) system sequence diagram with a well-defined semantics. We can also formally prove the semantic equivalence of several instructions, even for non-deterministic instructions. The class of instruction languages provides a fruitful similarity between the structuring mechanisms for modeling business processes, textual system sequence diagrams, and programming languages, among others. This will ease the translation towards an implementation in a software system.

Graphical system sequence diagrams (SSDs), used in Requirements Engineering and Analysis, are hard to draw for larger use cases. However, we wanted to be able to express SSDs for such use cases as well. While we were designing a grammar for textual SSDs, we noted that we needed the same kind of structuring mechanisms as for modeling business processes and programming languages. Only the basic building block differs: e.g., an assignment statement for (imperative) programming languages and a (basic) interaction step for SSDs. Can we generalize this similarity idea, leading to (uniform) grammars for instruction languages in general, and exploit this? This paper presents a grammar for ‘structured’ Ws (where W stands for ‘Whatever’), leading to grammars for (textual) instruction languages. The general grammar contains constructs necessary for specifying all kinds of instructions. We illustrate our theory with examples from several different application areas. A basic building block might even be a picture or an icon, as we will point out. The main contribution of the paper is a powerful and uniform grammar for instruction languages in general, containing generic constructs for specifying various instructions conveniently. Exploiting the similarity between the structuring mechanisms for modeling business processes, (textual) SSDs, and programming languages will ease the translation towards an implementation in a software system. Usually the grammar rules for basic building blocks are domain specific and must be specified and added per application.