Applied Linked Data 2016-01-07

Attendees:
    * Steven Anderson (BPL)
    * Trey Pendragon (Princeton)
    * Anna Headley (CHF)
    * David Lacy (Villanova)
    * Corey Harper (NYU)
    
Linked Data Fragments Update:
    * David tried Trey's presentation code and spun it up. Worked.
Background:
    * Group's project was need for a way to cache remote vocabs for local use
    * LDF Spec implementation has potential to manage this
    * TRiple Pattern Fragments potentially help: http://linkeddatafragments.org/in-depth/#tpf
    *  2nd half of this work is how do we then use cached triples to do things like enrich URIs for use in SOLR & in UI
Converation about the current status of hosting Minted URIs (especially outside of Fedora).
CHF may be putting together a grant application that is related to our group's current interests.
    
Solr OCR Exclusion Update (?):
    * https://issues.apache.org/jira/browse/SOLR-3191 (looks like a December comment mentioned the patch no longer works)
    * Erik Hatcher reached out shortly before the Holiday, stating that he expected to rewrite the field feature over the break. 
Linked Data Usage Limits (Steven Anderson):
    * Library of Congress only seems to allow one hit per 3 seconds (http://id.loc.gov/robots.txt).
    * In my case, would query them to get a search result for a term and then do 20 asynchrous requests to resolve each returned result (altLabels, broader, narrower mostly). This ended up having the Library of Congress block me.
      * Their suggested solution is to either:
          * A) Not do broader / narrower (?)
          * B) Try to resolve the term directly as "http://id.loc.gov/authorities/subjects/label/[term].rdf" (So  http://id.loc.gov/authorities/subjects/label/Health.json which means exact match only)
  • relationship between caching / proxying - to get around restrictive terms of service on providers' side.
LDP Implementation of Versioning (Memento) (Not our work, but an update)
 - There's work going on for adding memento support to URIs in LDP.
Asserting Triples about External Resources + Versioning Them
- Need: not only cache URIs, but also make assertions about them. (e.g. the label is actually X)
- But probably only want to make an assertion against a specific version of a triple.
- Also probably want to record author of assertion
- This can be done with either named graphs or reification
- May be building this type of thing in Pheonix (not Rails) (http://www.phoenixframework.org/)
Two same concepts but in different vocabularies:
    * Tag twice? Use skos:exactMatch / skos:closeMatch? Etc.
    * It depends?
      * Probably just assert skos:exactMatch
    *  Historically you mint your own uri then skos:exactMatch or owl:sameAs to both terms.
      * Also can wait for code from the previous conversation to be available.
    * In a case that you can update the remote source (ie. Geonames), can add skos:exactMatch to other sources (ie. TGN).