Article Preview
TopIntroduction
According to well-known definitions (Davenport & Short, 1990; Rummler & Brache, 1995; Weske, 2007), a business process consists of a number of tasks designed to produce a product or service and is meant to cross functional boundaries in that it may involve members of different departments. This point of view has spurred a fruitful line of research, now labeled as activity-centric, and the development of a standard notation, BPMN (Business Process Model and Notation) (OMG-BPMN, 2011). In the activity-centric approach, the notion of business process is the primary one and is usually thought of as a number of tasks to be performed in a sequence established by means of precedence rules, which are based on the completion events of the tasks. The precedence rules form the control flow of the process. The data flow is based on process variables, which provide the inputs of the tasks and receive the results produced by them.
In recent years, a shift of interest from the activity-centric perspective to the artifact-centric one took place. The main criticism to the activity-centric approach is to view business processes as an extension to procedural programs, where the extension consists in allowing long-lived activities (e.g., human tasks) to take place. Low flexibility and no links with business entities are the major drawbacks.
On the contrary, the artifact-centric approach emphasizes the identification of the key business entities (called artifacts) and of their life cycles, which show how the artifacts evolve over time through the execution of business operations. The analysis of the operations of a global financing division resulted in a high-level model consisting of 3 major artifact types, whose life cycles include 18 states and approximately 65 tasks (Chao et al., 2009). The major benefit is the right level of granularity, which facilitates communication among the stakeholders and helps them focus on the primary purposes of the business.
In the artifact-centric approach, there are three major issues to cope with, i.e., structure, dynamics and coordination (Hull, 2008). Structure is about the properties (attributes and associations) of the artifacts involved, dynamics encompass the artifact life cycles and coordination is about harmonizing the various life cycles so as to achieve the purpose of the business process. These issues can be dealt with in various ways, ranging from separate models to holistic ones.
The contribution of this paper is to illustrate and to compare two forms of representation related to the issues of dynamics and coordination. In one form, all the life cycles involved in a business process appear in a single model, and, in the other, the life cycles are worked out in separate models; the resulting models are called compact models and compound ones, respectively. The overall structure is defined in an information model that includes the entity types along with their attributes and relationships. Terms entity and artifact will be often used interchangeably: the difference is that term artifact indicates the entities whose life cycles matter to the process.
The coordination of the life cycles is obtained by means of tasks affecting artifacts of different types: such tasks are called spanning tasks. Three kinds of coordination are addressed in this paper: aggregation (and disaggregation), synchronization and matching. A number of motivating examples are presented along with the notation used to define them. This notation is named ARTS (ARtifacts and TAsks) as it considers both artifacts and tasks as first-class citizens of business process models.
Models based on ARTS are conceptual models: in addition to helping analysts understand the problem under consideration, they facilitate communication between analysts and developers. In particular, tasks are described in terms of behavioral requirements, which consist of the effects on the entities affected and of the constraints that limit their execution. The effects and the constraints are expressed declaratively by means of post-conditions and pre-conditions, respectively; they are based on the structure of the entities provided by the information model.
This paper is organized as follows. First it addresses aggregation and disaggregation and then synchronization and matching. The last sections describe the related work and provide the conclusion.