International Journal of Software Engineering and Knowledge Engineering

This paper introduces an approach to modeling software project dynamics using an attributed graph grammar formalism. A project situation is modeled and presented to a "player" who is supposed to act as project manager of a simulated project. Events and internal processes of the simulated project are modeled using a custom-designed formalism that integrates attributed graph grammar concepts with a quantitative simulation mechanism resembling System Dynamics. Both ingredients were pragmatically selected and blended based on (1) their comparative simplicity, (2) their expressive power, and (3) their intuitive appeal for bridging the cognitive gap between real-world software engineering experiences and an executable simulation model.

This modeling formalism was applied in the SESAM educational software engineering game. The modeling formalism was supported by a family of graphical SESAM editors. Resulting models could be animated and served as simulated projects in the game runs. SESAM was applied in graduate courses on software project management at the University of Stuttgart. Experiences with the modeling approach are reported. The design history and design rationale that led to using graph grammars as the underlying formalism are described and discussed. The design of the SESAM modeling formalism is presented as an (involuntary) case study in adopting a graph grammar approach.

Software process dynamics challenge the capabilities of process-centered software engineering environments. Dynamic task nets represent evolving software processes by hierarchically organized nets of tasks which are connected by control, data, and feedback flows. Project managers operate on dynamic task nets in order to assess the current status of a project, trace its history, perform impact analysis, handle feedback, adapt the project plan to changed product structures, etc. Developers are supported through task agendas and provision of tools and documents. Chained tasks may be executed in parallel (simultaneous engineering), and cooperation is controlled through releases of document versions.

Dynamic task nets are formally specified by a programmed graph rewriting system. Operations on task nets are specified declaratively by graph rewrite rules at a high level of abstraction. Furthermore, editing, analysis, and execution steps on a dynamic task net, which may be interleaved seamlessly, are described in a uniform formalism.

The idea of a combined reference model- and view-based specification approach has been proposed recently in the software engineering community. In this paper we present a specification technique based on graph transformations which supports such a development approach. The use of graphs and graph transformations supports an intuitive understanding and an integration of static and dynamic aspects on a well-defined semantical base. On this background, formal notions of view and view relation are developed and the behaviour of views is described by a loose semantics. The integration of two views derived from a common reference model is done in two steps. First, dependencies between the views which are not given by the reference model are determined, and the reference model is extended appropriately. This is the task of a model manager. If the two views and the reference model are consistent, the actual view integration can be performed automatically. For the case of more than two views more general scenarios are developed and discussed. All concepts and results are illustrated at the well-known example of a banking system.

In this paper, we introduce a new concept for the structured development of rule-based graph transformation systems that allows to build up large systems from small pieces in a modular way. The concept is called nested graph transformation unit and consists of a network of local graph transformation units. While each local unit provides a set of rules and control conditions (including input and output descriptions), the net reflects the import structure of the system where each local unit can use its direct neighbours without any restrictions. In particular, the import structure may contain cycles. It is shown that nested graph transformation units have a reasonable operational semantics by iterating the so-called interleaving operator, which interleaves local rule applications with calls of used units according to the local control conditions. It turns out that the iterated interleaving semantics is the smallest fixed point of the interleaving operator provided that the involved control conditions are continuous in a certain sense.

In this paper we introduce a formal object specification technique. We call it OST in analogy to the well-known object modelling technique (OMT), because it adapts the OMT methodology. In order to have a better integration of object and functional model we propose to specify the operations of the functional model in terms of instance diagrams in a rule-based way as it is known from graph transformations. Moreover OST allows complex data constraints in the object model concerning existence and uniqueness of objects and attributes in a specific context. These constraints are described graphically over instance diagrams and they can be ensured automatically within the functional model. The correctness of the mechanism for ensuring the constraints is guaranteed by the formal semantics of OST.

Entity-Relationship modelling is a visual, hence a rather intuitive technique for specifying the structure of complex data. The technique is popular in part because the structure of an ER-model is easily grasped, and it is usually supported by diagrams or other visualizing tools. This paper deals with a detailed analysis of ER-modelling with the goal of deriving an algebraic specification for a given ER-model. This is motivated by considerations regarding program specification for data intensive applications. We indicate how the technique demonstrated here may be combined with formal techniques for specifying the functional behavior of a system.

This study is based on the Structured Process Model formalism previously proposed by the same authors, and describes some techniques for scheduling an execution plan which conforms to the chosen process but adapts to all the potential changes in the value of specific variables of the project. The executive plan is generated by means of Incremental Refinements. Each increment is the consequence of one or more decisions taken by the manager.

All the contingencies which can occur during the execution of the project are classified in a finite number of types of events which form the basis for the definition of all the transition laws between the possible states of the process. Naturally, the manager's decisions, and therefore the composition of the increment, take into account the changes in the value of the variables of the process and of the process model itself.

In addition, the study proposes the handling of fuzzy activities, by means of which the problems of recycles and reprocessing are solved. These activities also provide a better management of the execution times of the project.