Knowledge Graph Construction to Facilitate Chemical Compound Hazard Assessment in the TOXIN Project

Student Name
Guillaume Vrijens
Thesis Type
Master Thesis
Thesis Status
Academic Year
2022 - 2023
Master in Computer Science
Christophe Debruyne

his master thesis presents a method for integrating multiple data sources from the field of toxicology into a knowledge graph and linking it with the TOXIN knowledge graph to facilitate the hazard assessment of new compounds. The proposed method uses a hybrid approach, combining an ontology and Linked Data to capture the granularity of the toxicological domain and provide a consistent representation while maintaining the flexibility of Linked Data. The ontology used in the method is the ToXic Process Ontology (TXPO), which offers a structured and reliable representation of the relationships between toxicological processes. The method also incorporates the use of named graphs and provenance information to store different opinions on data and track the integration of different sources. The feasibility and utility of the proposed method for building the knowledge graph are demonstrated through the development of a prototype, the TOXIN enriched knowledge graph (TEKG). Finally, this project illustrates the potential value and usefulness of a knowledge graph such as TEKG for improving access to relevant information, offering a satisfactory representation of the toxicological domain and supporting domain-specific tagging mechanisms.