Description
Contrary to traditional process models, declarative process models define a set of declarative constraints to specify the behavior which a process should adhere to. In the scope of process mining, declarative process discovery aims to derive such constraint sets from event logs. Here, a problem for current discovery techniques is that of inconsistency. That is, dependent of certain event log characteristics, the derived constraint set may contain contradictory constraints. This in turn however makes the discovered model unusable, as contradictory constraints make it impossible to execute declarative process models, thus hampering previous process discovery efforts. In this work, we present an approach for resolving inconsistencies in declarative process models, based on methods from the scientific field of inconsistency measurement. We introduce our approach algorithm and evaluate its feasibility with data sets of the BPI Challenge 2017.
Resolving Inconsistencies in Declarative Process Models based on Culpability Measurement
Contrary to traditional process models, declarative process models define a set of declarative constraints to specify the behavior which a process should adhere to. In the scope of process mining, declarative process discovery aims to derive such constraint sets from event logs. Here, a problem for current discovery techniques is that of inconsistency. That is, dependent of certain event log characteristics, the derived constraint set may contain contradictory constraints. This in turn however makes the discovered model unusable, as contradictory constraints make it impossible to execute declarative process models, thus hampering previous process discovery efforts. In this work, we present an approach for resolving inconsistencies in declarative process models, based on methods from the scientific field of inconsistency measurement. We introduce our approach algorithm and evaluate its feasibility with data sets of the BPI Challenge 2017.