The Mining of Activity Dependence Relation based on Business Process Models

With the development of process recommendation, dynamic adaptation and automatic modeling, the requirement of explicit and formalized expression of activity dependence relation in the business domain is becoming more and more urgent. However, these relations more exist in the minds of domain experts or in the unstructured documents, which leads process modeling and adaptation are a time-consuming and error-prone process. To solve this problem, a relation mining method is proposed for obtaining activity dependence relations. The formal description of these relations is defined in control flow perspective, which is expressed as serial-dependence relations and parallel-dependence relations in the form of three tuples after analyzing all the control flow patterns. And a mining algorithm is proposed for mining these two types of relations based on the process model. The correctness and performance of this algorithm are verified by a large number of experiments, and the experimental results show this method can quickly and accurately extract all the activity dependence relations from the existing process models in a business domain.