Extending AgreementMakerLight to Perform Holistic Ontology Matching.

Creating rich knowledge graphs that allow the representation of data encompassing multiple domains requires the integration of different ontologies. However, the challenge of matching multiple ontologies is not properly addressed by the current pairwise strategy espoused by state-of-the-art ontology alignment systems. We have extended the ontology alignment system AgreementMakerLight (AML) to address this particular challenge through a scalable cluster-based incremental matching strategy. We make use of AML's fast and precise matching algorithms to determine the semantic affinity between the ontologies and cluster them, then apply AML's full ontology matching pipeline incrementally, within each cluster, by matching and then merging ontologies pairwise. The strategy was applied to the integration of 28 biomedical ontologies and achieved a runtime reduction of almost 50%. This poster expands on the extensions applied to the AML system as the technical contribution that accompanies our In-Use Technology accepted submission "Matching Multiple Ontologies to Build a Knowledge Graph for Personalized Medicine".