Propagating Ontology Changes to Declarative Mappings in Construction of Knowledge Graphs

Knowledge Graphs (KGs) are usually constructed through a set of data transformation pipelines that turn heterogeneous sources into triples following a set of rules. These rules, usually in the form of mapping rules (e.g., RML, R2RML, etc.), are a key resource for the construction of the KG as they describe the relationship between the input data sources and the ontology terms. Several efforts have been made to manage and describe the evolution of the ontology; however, its propagation over interrelated assets (e.g., mapping rules) is commonly done in manual processes. In this paper, we present a preliminary approach to automatically project the evolution of the ontology on the mapping rules used to construct the KG. For each potential change, we analyse the impact on the mappings and the required steps to ensure that the KG is up-to-date w.r.t. the ontology. We implement our solution in fully declarative workflows and demonstrate its benefits in a real-world project in the public procurement domain.

keywords: Knowledge Graphs, Ontology Evolution, Mapping Rules, Impact Assessment