Process Performance indicators (PPIs) play an important role in monitoring the performance of operational procedures. Both defining and measuring suitable
PPIs are key tasks for aligning strategic business objectives with the operational implementation of a process. A major challenge in this regard is that perspectives on the same real-world phenomenon differ among the stakeholders that are involved in these tasks. Since the formulation of PPIs is typically a managerial concern, there is a risk that these do not match with the exact operational and technical characteristics of business processes. To bridge this gap, the concepts
described in PPIs must first be linked to their corresponding process elements. Establishing these links is paramount for the monitoring of process performance.
Without them, the values of PPIs cannot be computed automatically. However, the necessary links must currently be established manually. A task which is tedious and error-prone, due to the aforementioned incoherence between the different perspectives. The goal of our work is to overcome the efforts involved in the manual creation of links by automating this step. To achieve this, we developed an approach that automatically aligns textual PPI descriptions to the relevant parts of a process model. The approach takes a textual PPI description and a process model to which the PPI relates as input. Given this input, the approach generates an alignment in three steps. (1) Type classification: We make use of a decision tree classifier to identify the type of a given PPI, which is important because it affects the number and kinds of process model elements that should be aligned to a PPI. (2) PPI parsing: We parse the textual PPI description to extract those phrases that relate to specific parts of a process, making use of natural language processing techniques. (3) Alignment to process model: Finally, given the identified measure type and the extracted phrases, we compute an alignment between the phrases and the process model. A quantitative evaluation with a set of 173 PPIs obtained from industry and reference frameworks, demonstrates that our automated approach produces satisfying results.