English

Abstract

Business process mining is a solution to discover business processes. These techniques take event logs recorded by process-aware information systems. Unfortunately, there are many traditional systems without mechanisms for events collection. Techniques for collecting events (which represent the execution of business activities) from non-process-aware systems were proposed to enable the application of process mining to traditional systems. Since business processes supported by traditional systems are implicit, correlating events into their execution instances constitutes a challenge. This paper adapts a previous correlation algorithm and incorporates it into a technique for obtaining event logs from traditional systems. This technique instruments source code to collect events with some additional information. The algorithm is applied to the events dataset to discover the best correlation conditions. Event logs are built using such conditions. The technique is validated with case study, which demonstrates its suitability to discover the correlation set and obtain well-formed event logs.