Declarative semantics of actions and instructions

Abstract

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.