Fuzzy traceability: using domain knowledge information to estimate the followed route of process instances in non-exhaustive monitoring environments

In the frame of the process monitoring domain, traceability in environments with non-exhaustive event logs remains as a challenge. In these scenarios, the discovery of the route followed by a particular item along the entire process is a difficult task since it is not possible to access all the required pieces of information from processes under monitorization. To tackle this issue, the concept of fuzzy traceability is brought into the scene. The gist of the latter is to use contextual information derived from the domain of interest itself to infer the most probable route followed. To carry out this task, the proposed algorithm takes advantage of additional sources of machine readable information that describes in a more detailed manner the process models under study. This information is included in the process models using the advanced features supported by the BPMN-E2 (Business Process Model and Notation - Enhanced Expressiveness) specification, an extension of the well-known BPMN notation. In this way , it is possible to properly use as inputs: time restrictions of the activities included in the process; decision-making and monitoring points included; and the effects derived from the activities undergone. As a consequence, a probabilistic estimation of the route followed is generated by combining this information according to the presented algorithm. After the validation and simulation of the fuzzy traceability algorithm using real-world models, the results obtained are positive and show that, as the contextual information included grows, the route estimation gets more acurated. This high success rate suggests that the fuzzy traceability proposal is useful for the analysis of processes with poor quality of monitoring information, and outdoes the application of more conventional traceability techniques.