Some human-in-the-loop feedback loops within the Rich Context knowledge graph.
Apply manually curated metadata to override the stream generated using discovery APIs, then generate a new corpus for the ML competition.
Generate metadata to send to RePEc for their partnership on an HITL feedback with authors confirming the metadata inferred by ML models.
Glean the publication <=> dataset links that we can from the corpus for the first RCC competition.