anilpacaci / cord19-kg-importer Goto Github PK

Extracted relations refer to ~100K publications from PMC whereas the metadata is available only for ~22K. We ca retrieve the metadata from

• via PMC APIs
• other similar datasets

Starting from a search result in covidex.ai, explore other facts that are extracted from the same document.

Following three relationships:

• Chem-disease pairs in the context of publication
• Gene-disease pairs in the context of publication
• Chem-gene pairs in the context of organism, publication
  • different publication might extract different interactions

Related to #5
Current KG does not have metadata for all articles referred. It is necessary to create empty nodes only with ID until we extract metadata from other sources.

It can be done by simply extracting pmid fields from relationship data.

This is related to issue #4
KG schema should represent new fact nodes that represent hyperedges, i.e., the extraction context for relationships