Discovering Instance-Spanning Constraints from Process Execution Logs Based on Classification Techniques

Abstract

Process-aware Information Systems (PAIS) have become ubiquitous in companies. Thus the amount of data that can be used to analyze and monitor process executions is vast. The event logs generated by PAIS might contain information about decision making processes and can support the understanding and improving of procedures in companies. Mining decisions and constraints from logs has already been investigated, but so far only for each instance in a separate manner. However, in many practical settings instances are connected to each other if they share, for example, the same resources. Therefore, we present an approach for discovering Instance-Spanning Constraints (ISC) from event logs. The main idea is to identify instance-spanning attributes in the logs and to separate the logs accordingly. Based on these projections, classification algorithms are applied in order to obtain ISC candidates. The feasibility and applicability of the approach is evaluated based on artificial as well as real-life logs. The discovered ISC candidates are then assessed by domain experts.