From Declarative to Imperative UML/OCL Operation Specifications

Abstract

An information system maintains a representation of the state of the domain in its Information Base (IB). The state of the IB changes due to the execution of the operations defined in the behavioral schema. There are two different approaches for specifying the effect of an operation: the imperative and the declarative approaches. In conceptual modeling, the declarative approach is preferable since it allows a more abstract and concise definition of the operation effect and conceals all implementation issues. Nevertheless, in order to execute the conceptual schema, declarative specifications must be transformed into equivalent imperative ones.Unfortunately, declarative specifications may be non-deterministic. This implies that there may be several equivalent imperative versions for the same declarative specification, which hampers the transformation process. The main goal of this paper is to provide a pattern-based translation method between both specification approaches. To facilitate the translation we propose some heuristics that improve the precision of declarative specifications and help avoid non-determinism in the translation process.