Causal Process Mining from Relational Databases with Domain Knowledge

Abstract

The plethora of algorithms in the research field of process mining builds on directly-follows relations. Even though various improvements have been made in the last decade, there are serious weaknesses of these relationships. Once events associated with different objects that relate with a cardinality of 1:N and N:M to each other, techniques based on directly-follows relations produce spurious relations, self-loops, and back-jumps. This is due to the fact that event sequence as described in classical event logs differs from event causation. In this paper, we address the research problem of representing the causal structure of process-related event data. To this end, we develop a new approach called Causal Process Mining. This approach renounces the use of flat event logs and considers relational databases of event data as an input. More specifically, we transform the relational data structures based on the Causal Process Template into what we call Causal Event Graph. We evaluate our approach and compare its outputs with techniques based on directlyfollows relations in a case study with an European food production company. Our results demonstrate that directly-follows miners produce a large number of spurious relationships, which our approach captures correctly.

Process Mining over Multiple Behavioral Dimensions with Event Knowledge Graphs

“…We need automated techniques to infer relevant relations that take the temporal semantics of the df-relationship into account. Promising first steps are techniques that explicitly allow to incorporate domain knowledge when inferring causal relationships from relational data [56], or use ontologies [6,7] for extraction. Specifically, dynamically changing relationships and changes of object properties [39,40] still need to be considered.…”

“…Alternative formalizations of Definition 8 define just a partial order over events [4,30,55,56] describing the local directly-follows relation wrt. various entities 6.…”

“…All steps of the method can be implemented as a series of Cypher queries 6 to construct event knowledge graphs in a graph database for our running example [28] as well as for various real-life datasets comprising single and multiple event tables [24]; several event knowledge graphs of real-life processes are available [19][20][21][22][23][24]. A variant of event knowledge graphs, called causal event graph that only models events but not the entities, can be extracted automatically from relational databases [56].…”

Clustering Object-Centric Event Logs

“…Each row shows an event that has an identifier, a timestamp, an activity, related objects, and attributes. objects interact with each other in a process [3,4,11,12,27], for example, considering a Purchase-to-Pay (P2P) process where orders, items, and customers are involved [14,13]. Several process discovery techniques have been developed on top of event logs with multiple case notions [2,8,17,19,20,21].…”