SciLake's scientific knowledge graphs are built using semi-supervised tools and methods for interlinking data (ontology, entity, relationship, property). Based on recent work on graph generating dependencies (GGD), these methods will generate explainable fuzzy logical rules that capture expert knowledge and ML-driven measures of graph similarity. Rules based on GGD are used to resolve entities and relationships, to verify and enforce key values and to generate further knowledge based on expert or ML models.
