Connecting food ontologies and graph databases (MSc/BSc)
At FrieslandCampina R&D, we deal with large amounts of diverse data and do analytics on them, particularly about food characteristics. Property graph databases (mainly Neo4J) are a great tool in nowadays analytics infrastructures. However, integrating them with existing knowledge graphs / ontologies is still difficult. This work deals with approaches on how to bring ontologies (e.g. FoodON) efficiently and sustainably into a graph database (e.g. Neo4J) such that it can be easily integrated with other datasets. Questions around best practices, most relevant ontologies and possibly new technologies which are needed should be explored.
We believe that we can utilize the vast amount data on food characteristics to improve our innovation robustness, speed and impact. To do so, however, data needs to be placed in context to be easier understandable for humans and algorithms. Within this context we are looking at a student that can help us to bridge the gap between real world and data world using graph databases, knowledge graphs and ontologies to bring data into the right context.
To do so, we are interested in exploring the interface between RDF / knowledge graphs and the property graph world. We believe that by leveraging the best of both worlds we can enrich our internal datasets with external information easier and faster. Within this internship, we expect a successful candidate to scout for approaches to integrate RDF datasets into a propertygraph (Neo4J) and apply this to the FoodOn ontology such that we can map our food products to the relevant entities in FoodOn.
- Review previous work on the integration and interoperability between ontologies and graph databases, including technologies, relevant ontologies, best practices, challenges
- Design and implement a solution to connect an ontology (e.g. FoodON) to the Neo4J graph database
- Evaluate the effectiveness of the solution using a real case study at FrieslandCampina
The work in this thesis entails:
- To collect full-text articles or PDFs relevant to selecting ontologies and integration strategies between property graphs and RDF knowledge graphs
- Design a data ingestion from RDF / ontology to Neo4J and an evaluation thereof using relevant methodologies
- To design an IoT-based smart farming solution using an existing low-code platform, such as NodeRed or ThingML
- To deploy, demonstrate and validate the low-code solution in the NetLogo simulation environment
- FoodOn: A farm to fork ontology, https://foodon.org/
- Neo4j graph database, https://neo4j.com/
- Courses: INF-33306 Linked Data and INF-22306 Programming in Python not formally required, but useful
- Required skills/knowledge: RDF / ontologies / semantic web, programming in Python or similar language, basics of databases, graph database skills are nice to have, otherwise to be acquired during the thesis
Key words: Knowledge graph, ontology, RDF, property graph, graph database, semantics, food
- Önder Babur (email@example.com)
- Carsten Ersch (Carsten.Ersch@frieslandcampina.com)