A Context Driven Approach for Workflow Mining

Our approach analyzes the data dependencies in the trace to discover the context of the actions that appear in the trace. Existing work on workflow mining ignores the Using the context information we can decide whether the two dataflow aspect of the problem. This is not acceptable occurrences of the same action correspond to the same node for service-oriented applications that use Web in the workflow or not. As a result, unlike the previous work services with typed inputs and outputs. We propose [van der Aalst et al., 2004; Cook and Wolf, 1998b; Agrawal a novel algorithm WIT (Workflow Inference from et al., 1998], we are able to learn workflows with non-unique Traces) which identifies the context similarities of action nodes. Furthermore, the context discovery can easily the observed actions based on the dataflow and uses be generalized to work with causal dependencies instead of model merging techniques to generalize the control data dependencies. Thus, the ideas presented in this work flow and the dataflow simultaneously. We identify can be applied to other areas such as learning domain specific the class of workflows that WIT can learn correctly.