AN INCREMENTAL PROCESS MINING ALGORITHM

A number of process mining algorithms have already been proposed to extract knowledge from application execution logs. This knowledge includes the business process itself as well as business rules, and organizational structure aspects, such as actors and roles. However, existent algorithms for extracting business processes neither scale very well when using larger datasets, nor support incremental mining of logs. Process mining can benefit from an incremental mining strategy especially when the information system source code is logically complex, requiring a large dataset of logs in order for the mining algorithm to discover and present its complete business process behavior. Incremental process mining can also pay off when it is necessary to extract the complete business process model gradually by extracting partial models in a first step and integrating them into a complete model in a final step. This paper presents an incremental algorithm for mining business processes. The new algorithm enables the update as well as the enlargement, and improvement of a partial process model as new log records are added to the log file. In this way, processing time can be significantly reduced since only new event traces are processed rather than the complete log data.