در روابط بین مدل های جریان کار و انواع سند

The best practice in information system development is to model the business processes that have to be supported and the database of the information system separately. This is inefficient because they are closely related. Therefore we present a framework in which it is possible to derive one from the other. To this end we introduce a special class of Petri nets, called Jackson nets, to model the business processes, and a document type, called Jackson types, to model the database. We show that there is a one-to-one correspondence between Jackson nets and Jackson types. We illustrate the use of the framework by an example.

Data modeling and process modeling are two essential activities in requirements analysis and design of information systems. They are using different techniques and normally they are performed independently. Since both techniques are defining essential aspects of an information system they have to be integrated at some point in the development process, but normally this is at the level of programming. In this paper we show that data modeling and process modeling can go hand in hand from the beginning of the development process of so-called case-based information systems. The characteristic of these information systems is that they are developed to support the handling of cases, such as the treatment of a patient, the handling of an order or the delivery of a service. For each case type there is a workflow defining the tasks to be performed for the case. A workflow is a process with a clearly defined start and end state. In this paper we use a special class of Petri nets to model workflows, the so-called workflow nets [1]. Since in each task of the workflow something happens to the case, it is to be expected that the data type to record the case data is related to the structure of the workflow. The case data is recorded in the case document, the structure of which is a document type. We show that if we restrict ourselves to a special class of workflow nets, the so-called Jackson nets, then there is a tree shaped document type for the case data, called the Jackson type, that contains the same information as the workflow net. One of the main results in this paper is that there is a one-to-one correspondence between the document type and the workflow description, so from one we can derive the other. This is similar to the classical program design method of Jackson [2] which is the reason we called the workflow nets and the document types after this author. The organization of the rest of this paper is as follows. In Section 2 we give the system development context for our work and we give a motivating example. In Section 3 we introduce Jackson types. In Section 4 we introduce the Jackson nets and in Section 5 we study the relationships between Jackson nets and Jackson types. In particular we prove that if two Jackson nets are derived from the same Jackson type they are isomorphic and that if it is possible to derive the same Jackson net from two different Jackson types, these types are algebraical equivalent. In Section 6 we continue with the motivating example. Here we show how we can derive an XML document type from a Jackson net and demonstrate how it provides a logical structure that helps the user to formulate queries over the cases of the workflow. In Section 7 we discuss the usefulness of Jackson types. Finally, we discuss related work in Section 8. Theconclusion of the paper is given in Section 9.

In this paper we have presented a framework in which it is possible to establish a straightforward relationship between the process model and the data model of a case-based information system such as a workflow system. To this end we introduced a special class of Petri nets, called Jackson nets, to model the business processes, and a document type, called Jackson types, to represent the data model. We have shown that there is a one-to-one correspondence between Jackson nets and Jackson types with several interesting theoretical properties. Finally, we have illustrated the use of the framework by an example in which it is shown that the resulting data model allows the straightforward formulation of queries over runs of the system. In future research we intend to extend the presented framework to address the problem of constructing and integrating the different data models that are associated with each event in a run of the system. Some preliminary work on this was presented in [22] but no attempt was made yet to integrate it into the framework that is presented here. In addition we will develop tools to support the design of workflows based on Jackson types.