Making triples

One of the goals of the Civil War pilot is to “create some metadata as linked data”.  This has been  challenging, because we weren’t able to find (or make) a “linked data making tool” that would be really helpful.  (We investigated Protege, but it seemed more designed to manage ontologies than to help you create “instances” using those ontologies.)   This seems consistent with the difficulty of finding  good general-purpose metadata creation tools that anyone could use, for creating metadata using, e.g.,  XML schemas, outside of a particular suite of (usually proprietary) software.

As it became apparent we weren’t going to have time to build something ourselves, integrated with Sesame, we started creating just very simple triples in N3 in Notepad.   Something like:

@prefix dcterms: <http://purl.org/dc/terms/121>.

<http://oururlforsome/resource/121>  dcterms:date “1864-07-22”.

Meaning – “this resource (which happens to have 121 at end of its URI), has a property of  date  (in the Dublin Core Terms ontology/definition) of July 22 1864” —  not as semantically rich as it could be, but we can get to that later.

Then we use the RDF Validator at http://www.rdfabout.com/demo/validator/index.xpd created by Joshua Tauberer, to validate our N3, before using the Sesame OpenRDF Workbench “cut and paste” to enter them into our graph store.

Really rudimentary, but taught us some things:

• We need a tool to do this.  It’s not as bad as trying to create raw MARC, but I find RDF  harder to “human read” than some metadata using XML schemas, especially if the URLs are not expressive of the meaning.
• The tool should take care of syntax if possible, like making sure each triple has a period at the end.  😉
• It should be possible to have a standard list of vocabularies and select predicates from lists…  by auto-complete or some other method.  It could also be good to have some “required” relations if we want that…  other kinds of logical validations and visualizing how the added data fits in the graph would be nice too.
• Being able to drag URLs from anywhere in a browser into a form would be great.
• However, doing it by hand at the beginning does really cause you to think about what you’re doing in a different way.  You’re constantly asking yourself “What am I describing, what am I trying to say about it, and what’s the relationship between this and that?”.

So, these observations will probably find a place our discussions of what we want to tackle  next and what we will need to really make linked data useful for us.