Page History

• Metathesaurus -
  
  • John working on Stardog relational mapping to UMLS RDB
    • Get DB specs and look at loading locally at NCI
    • Loading the database took half hour, but included all tables.
  • Option - use LexEVS loaded RDB and build Elastic Search on that?
• Hosted Mappings
  • Map sharing - for review purposes and/or publication purposes
  • Common format for mapping to enable sharing
  • Don't limit future options by current capabilities
  • Kim thinks building a single ttl file with all mappings would work.  Then load this to triplestore.  Should not be reliant on whether vocabulary is present in triplestore.
    • If mapped vocab is present it should resolve
    • Need same info in ttl as we currently have in mapping - extract from LexEVS (source, score, rank, target, etc)
• Hosted Value Sets
  • John already working on this
• Coding scheme metadata - copyright, contacts, explanation info. 

• Search knowledge
• Ontology knowledge

Centralizing data sources is generally a good idea.  Need to figure out what users really need. Who are the actual users?  How do we prototype in a way to get user feedback? How to keep people from adopting prototype as the final product? 

John is looking at our current LexEVS functionality and trying to replicate.  But the use case may be more simple than that.  Give me my data and make it fast.  We have a very limited user set right now.  Who are the representative users for the GDC and how do we get their input? 

We NEED to know the requirements and the users.  Build a minimal tool, put it out, accept feedback, expand from there.  Understand when the tool is due.

Logs have no errors but the performance is poor.  Try and debug. Since it is outside firewall, perhaps there is some security slowing things down? 

Scott says could be memory mapping - should check installation and configuration.