Applied Linked Data Call 2015-11-12

Attendees:

Linked Data Fragments Update:

  • Steven still working on the Repository interface on a branch.
  • ActiveTriples still doesn't have any momentum for a 0.8 release yet. So still waiting on that before attempting integration with it.

Atomic Updates Update:

  •  Rough update notes in the etherpad at: https://etherpad.wikimedia.org/p/Hydra-LDP-20151112
    • Of note, using "fl" to select stored fields with only a certain suffix is indeed possible. Essentially a: "fl=id,*_ssim" at the end of your Solr query will only pull the "id" and "_ssim" fields.
    • However, it would be much more clean to just exclude the OCR text we don't want returned. As such, there is the following JIRA issue: https://issues.apache.org/jira/browse/SOLR-3191
      • Recommend we test it out and see where coverage gaps may exist.
      • Current theory would be that the Hydra head specifically specifies the OCR fields to exclude rather than using a wildcard. This is because:
        • The patch may not work to exclude a wildcard pattern (would need to be tested).
        • No good suggestion yet on what the suffix/prefix might be that wouldn't affect Hydra's functionality. Needs some testing.

Metadata Enrichment Interface:

  • Steven did some hacking on an interface that goes beyond just autocomplete. In essence: you have a text field that you can enter a URI into if you know it. Otherwise, you can click the "lookup" button to bring up an interface to find a uri. It shows the broader and narrower terms for your results along with the usage of that term in the system in parenthesis. Clicking on any term (broader, narrower, actual) brings up the page from the remote URI for further details. Currently runs in Sufia 6.3.0 and will be made a generic separate gem.
    • Picture: metadata_enrichment_interface.jpg
    • Will likely change the clicking on the link functionality. For access to the remote web resource, likely use a small globe icon next to the text of each result.
       
  • Feedback:
    • Additional things that would be useful to show:
      • URI
      • Description (if available)
      • Alternative Labels for the actual term (in italics possibly?)
    • Other pieces of functionality:
      • Add ability to page through results.
      • Support for multiple authorities (likely via drop down for the soruce? Or potentially by combining results but would be tricky...).
      • API Interface for the broader and narrower results as not all vocabularies will use skos.
      • Clicking on the term could refocus the form to use that value and redisplay the results.
      • Version 2.0: Keep track of what they searched for and what they ended up picking.

Minting Linked Data URIs / Hosting Linked Data Vocabulary:

  • Oregon Digital has the Controlled Vocabulary Manager that does this and stores it right now in Marmotta.
    • It sort of works.
    • All it needs a link and the content to negotiate. 
    • Uses github by default for authentication / authorization.
    • Stores in Marmotta
      • As reliable as the postgres DB that backs it.
      • If you backup the postgres DB, then you have backed up Marmotta (and subsequently your linked data vocabulary).
    • Link: https://github.com/OregonDigital/ControlledVocabularyManager
       
  • You could also just embed RDFa in one of your pages. Same was the DC Pages content negotiate.
    • Literally the HTML representation of linked data.

Linked Data Fragments Standup:

  • Will be on Friday, November 20th at 11:00 AM PST / 2:00 PM EST on the same Google Hangouts link.

Next Official Meeting:

  • Next meeting will be December 10th at 9:00 AM PST / Noon EST.