Rule Mining in Action: The RuM Toolkit

Procedural process modeling languages can be difficult to use for process mining in cases where the process recorded in the event log is unpredictable and has a high number of different branches and exceptions. In these cases, declarative process modeling languages such as DECLARE are more suitable. Declarative languages do not aim at modeling the end-to-end process step by step, but constrain the behavior of the process using rules thus allowing for more variability in the process model yet keeping it compact. Although there are several commercial and academic process mining tools available based on procedural models, there are currently no comparable tools for working with declarative models. In this paper, we present RuM, an accessible and easy-to-use rule mining toolkit integrating multiple DECLARE-based process mining methods into a single unified application. RuM implements process mining techniques based on Multi-Perspective DECLARE, namely the extension of DECLARE supporting data constraints together with controlflow constraints. In particular, RuM includes support for process discovery, conformance checking, log generation and monitoring as well as a model editor. The application has been evaluated by conducting a qualitative user evaluation with eight process analysts.