Business process management is the key to aligning a company’s business with the needs of clients.
Companies increasingly rely on technology (for example RFID) to monitor, analyze, and improve their business processes. However, no technology is perfect, and quality issues often prevent us from using technology to its full potential.
In the SERAMIS project, we face challenges of low data quality with missing RFID reads, as well as false positive reads. False positive reads occur when an RFID reader detects an RFID tag that is in its signal range, but not meant to be detected. For example, a tag can be read through a wall while doing inventory checks.
Most business processes in the retail industry follow regular patterns. These patterns can be learned from historical executions, such that we can detect deviations in the current process execution.
Based on this insight, we developed an approach that helps to detect anomalies in the timestamps of RFID readings that happen during process execution. The approach is relying on historical observations of usual behavior and uses a stochastic model to find outliers in current records. It helps process analysts to identify and filter errors, and also to distinguish between measurement errors (where only one timestamp is an outlier) and real delays (where succeeding activities are also shifted).
The approach is described in the paper Temporal Anomaly Detection in Business Processes and is presented at the 12th international conference on Business Process Management (BPM), which is the “most prestigious forum for researchers and practitioners in the field of Business Process Management”cite.