Exploiting the Graph Structure of the Event Data Contained in SAP


Exploiting the Graph Structure of the Event Data Contained in SAP

Student: Roman Kassner

Title: Exploiting the Graph Structure of the Event Data Contained in SAP

Supervisor: Alessandro Berti

1st Examiner: Prof. Dr. Wil van der Aalst

2nd Examiner: Prof. Dr. Ulrik Schroeder


Process Mining techniques have proven helpful in gaining valuable data insights and thus
have rapidly gained importance and popularity over the last decade. Due to much data
being generated through SAP ERP systems, process mining techniques are often applied
to this data to yield insights about the underlying processes. However, this is done chiefly
using relational databases and, thus, relational queries. As investigated by several papers,
deviating from this approach and focusing on non-relational databases can yield some
benefits. Despite this, there has yet to be a further focus on non-relational databases in
process mining. Therefore, this paper will further explore how non-relational databases
can be more efficient than traditional relational databases in process mining. Mainly,
we will focus on exploiting the underlying graph structure of the data in the event logs
yielded by SAP systems and discover object-to-object relationships in an object-centric
event log. We will first present a tool to convert SAP data and an object-centric event
log into a graph in a non-relational database, specifically Neo4j. Secondly, we will give
several queries that use the underlying data’s graph structure and help gain insights and
improvement suggestions about the underlying processes. Lastly, we will compare these
queries with their relational counterparts and show how non-relational queries and already
implemented graph algorithms offered by Neo4j can yield significant improvements while
analyzing processes and the underlying data generated through SAP. By showing this, we
enable future research into this topic to use these insights and thus give more attention to
non-relational process mining approaches.



Alessandro Berti

Software Engineer


+49 241 80 21912