DGLink

DGLink introduces semantic interoperability within data portals through automated metadata extraction and knowledge graph construction. DGLink connects data semantically within a data portal and maps data to external knowledge to enable knowledge-driven data interpretation and discovery.

Structure

1. dglink/kg_construction - for programmatically building the knowledge graph from Synapse data.
2. dglink/example - reconstruct the basic example, we started with.
3. dglink/semantic_search - bring up web UI for semantic searching on the KG (see instructions bellow)
4. dglink/applications/project_similarity - Leverage KG embedding methods for accessing project similarity (need to install additional dependencies with) uv pip install ".[graph_embedding]"
5. dglink/resources - data and code for brining up the Neo4j instance.

Steps for running semantic search UI

1. Go into the semantic search directory with cd dglink/semantic_search
2. Bring up the service (and build images if required with) docker-compose up --build
3. Connect to the services.
   • semantic search UI http://localhost:5000/
   • neo4j browser http://localhost:7474/browser/