Linked data/RDF/Semantic web

From CURATEcamp
Jump to: navigation, search

Notes from CurateCAMP SE 2012 breakout session

Emory has a linked data interest group:

  • trains folks in basic SPARQL, sesame, SKOS
  • email list
  • pilot project will pull data out of silos and demonstrate potential (3 months)

RDF "readiness" cited as a first step.

OpenGraph (on Facebook) will make linked data more prevalent and personal.

Civil War regiments and SPARQL queries:

  • what can a SPARQL query tell you that SOLR cannot?
  • traversing relationships:
    • combine parks service data on regiment movements with institutional archives data on soldier diaries.
    • find indirect relationships between historical figures:
      • shared a regiment
      • in same battle
      • served under same officer

Questions

  1. What should we (library and associated personnel) be doing to prep for linked data, semantic web, and RDF?
  2. What audiences would have the most to initially contribute?
  3. Where does the "same as" relationship sit?
  4. Where are the hubs?
  5. Who hosts?
  6. Who is "trustworthy"?

Related Software

  1. Sesame (triple-store?)
    • OpenRDF Workbench (built on Sesame)
  2. Linked Sailor (?)
  3. Fact-find
  4. Zotero (as LOD-LAM project)
  5. FoF (Friend of Friend)
    • Early RDF example
  6. Open Graph

Notes

  • "Linked data is more about generic linking of data (forget RDF). Enabling graph of currently available data. RF is a way of serializing this data."

Obstacles to LD/SW/RDF implementation

  1. Developer time
  2. Narrow enough scope to manage a pilot
  3. Wide enough scope to be useful (large data)
  4. Scalability
  5. A challenge is demonstrating usefulness
  6. Have to ask questions that you can't ask of a faceting search engine.
  7. Difficult to present a "wow!" factor.
  8. Trust
    • Berners-Lee recommends assigned URIs.`