Fig. 2.

High-level Dug architecture. Dug makes study metadata searchable by parsing heterogenous metadata formats into a common format (ingest), annotating metadata using NLP tools to extract ontology identifiers from prose text (NLP annotation), searching for relevant connections in federated knowledge graphs using translator query language (TranQL) (concept expansion) and indexing this information into an Elasticsearch index. Dug’s web portal utilizes a flexible API to query and display search results back to end users