DigiEMine: Towards Leveraging Decision Mining and Context Data for Quality Control

Quality control processes in manufacturing often still rely on manual tasks. Applying decision mining can support users by providing valuable insight into the process. This paper discusses the potential of integrating contextual information into decision mining to achieve accurate and meaningful decision rules in the context of a case study stemming from the manufacturing domain. To explore this, a new approach, DigiEMine, is presented, which addresses the gap between information extraction and practical decision mining applications by integrating information extracted from engineering drawings with time sequence data in the form of diameter measurements of workpieces. The discovery of relational decision rules is enabled, allowing for contextualization of the decision rules. The output of this approach is presented in both textual decision rules and visually on engineering drawings, empowering users to make informed quality control decisions. The case study includes three datasets originating from cylindrical workpiece production. Results demonstrate the feasibility of the approach and the ability to generate meaningful decision rules across the tested datasets. Its potential applicability extends beyond the presented case study, with conceivable scenarios in multiple domains, such as healthcare or logistics, where integrating context information, such as regulatory data, with time sequence data is required to provide additional context for decisions.