The process mining field recently has adopted a graph-based approach for modeling and reasoning over event data using labeled property graphs.
From a process mining perspective, events are related to the various objects and entities involved in a process, providing a more realistic description of actual process dynamics in relation to domain concepts. From knowledge graph perspective, the approach extends knowledge graphs with a temporal perspective allowing to track behavior and changes of objects and entities over time, see:
- https://link.springer.com/article/10.1007/s13740-021-00122-1
- https://multiprocessmining.org/2021/11/02/how-do-event-graphs-help-analyzing-event-data-over-multiple-entities/
- https://github.com/multi-dimensional-process-mining/eventgraph_tutorial
Using inference rules over graph-based event data, we can address a number of very advanced use cases from inferring missing information (e.g. handling missing data) to inferring higher-level process concepts. All these techniques have been implemented using Labeled Property Graphs and Cypher queries.
The focus of this project is to explore and transfer modeling and inference over event data in knowledge graphs using RDF triple stores and SPARQL queries. Several Master projects are possible:
- Modeling Event Knowledge Graphs in RDF (concepts, ETL, querying, performance/scalability analysis)
- Event Knowledge Graphs and Semantic Ontologies (leveraging existing process-related ontologies for knowledge graph modeling and semantic inference/reasoning)
- Systematic development of semantic inference rules over event knowledge graphs in RDF
- AI-based querying and inference in Event Knowledge Graphs (exploring various AI techniques for retrieving information from event knowledge graphs and enriching event knowledge graphs)
- Other ideas proposed by students
Contact: Dirk Fahland (d.fahland@tue.nl)