A Knowledge-Based Representation of Business Process Models
Business process modeling technique is using by modern organizations to represent knowledge about their activities. Thus, collections of business process models designed by organizations may contain hundreds or even thousands models [1] represented using various modeling notations, from VAD (Value-added Chain Diagram) and IDEF0 used for high-level processes mapping to BPMN (Business Process Model and Notation), DFD (Data Flow Diagram), and EPC (Event-driven Process Chain) used to provide detailed process scenarios. Hence, the problem of storage, share, and reuse of organizational knowledge represented using business process models becomes relevant.
In this paper we propose usage of the knowledge representation model RDF (Resource Description Framework) to describe business process models, designed by an organization. The RDF model is based on “subject-predicate-object” statements (triples), which are convenient for machine processing. A set of these statements could be considered as a marked directed graph [2].
A set of classes and properties used to describe certain RDF knowledge representation model is called RDF Schema [2]. Proposed RDF Schema (Fig. 1), which was designed to provide business process models description, includes following classes and properties:
1) bpmodel:FlowObject — the class, which includes process flow objects, such as functions (bpmodel:Function), processes (bpmodel:Process), events (bpmodel:Event), and gateways (bpmodel:Gateway), associated with each other using bpmodel:triggers property;
2) bpmodel:OrganizationalUnit — the class, which includes organizational units, such as departments (bpmodel:Department) and roles (bpmodel:Role), associated with processes using bpmodel:executes property;
3) bpmodel:SupportingSystem — the class, which includes supporting IT-systems associated with processes using bpmodel:usedBy property;
4) bpmodel:BusinessObject — the class, which includes various business objects, such as material or information entities, associated with processes using bpmodel:isInputFor and bpmodel:isOutputOf properties;
5) bpmodel:KPI — the class, which includes Key Performance Indicators (KPI) associated with processes using bpmodel:measures property.
Activities related to a management of large collections of business process models are supported by the software, which is called business process model repository [1]. Thus, further research should consider usage of RDF storages both native and based on relational or NoSQL databases.
An RDF query language SPARQL is intended to retrieve and manipulate data stored in the considered RDF format. E.g., to retrieve information about organizational units and processes they execute, the following SPARQL query should be used (Fig. 2).
Hence, implementation of a repository, which contains business process models, described in the proposed manner, may allow organizations to store, share and retrieve knowledge about their activities for future reuse to design new or improve existing business processes. Considering a practice of models reuse, further research in this area should be focused on the development of similarity measure between RDF graphs, which represent business process models.
References:
1. Z. Yan, R. Dijkman, and P. Grefen, “Business process model repositories — Framework and survey”, Information and Software Technology, vol. 55, pp. 380–395, 2012. doi: 10.1016/j.infsof.2011.11.005.
2. RDF. [Online]. Available: https://www.w3.org/RDF/.
