Re: Inverse Properties in Microdata:, was Re: schema.org update, v1.8: added WebSite type; broadened isPartOf to relate CreativeWorks

On Aug 5, 2014, at 9:42 AM, Thad Guidry <thadguidry@gmail.com> wrote:

> DanBri,
> 
> Jarno does have a valid point about itemid lacking some documentation ... and I do not see examples or any mention on the getting started page: http://schema.org/docs/gs.html which probably should use at least 1 simple example of itemid usage on it.
> 
> BTW, looks like MedicalGuideline has mention of alternative use of itemid as an external enumeration:  https://github.com/rvguha/schemaorg/blob/master/data/examples.txt#L2919

I don't really see this as a different use, but "external enumeration" may be confusing. Basically, it's saying that when a well known identifier exists for an entity, that it is best to use it; this is a basic principle of Linked Data. In this case, NIH provides identifiers for all (most) drugs, so is a natural place to find such identifiers. Other places are DBpedia, Wikidata and Freebase. Many would like to mint their own identifiers (due to perceived issues of stability and control), which can be accommodated using sameAs to associate a local identifier with one or more external entities which are intended to be the same thing

IMO, the schema.org examples have long de-emphasized the usefulness of uniquely identifying items (nodes) in a graph, so that the use of @itemid remains confusing. I'd rather see more pervasive use of @itemid in such examples.

For example, the examples for WebPage really should use @itemid="", as that certainly identifies the page which is being described! Instead, the use of "structure" within a document is emphasized, which is then divorced of linked context.

Gregg

> -- 
> -Thad
> +ThadGuidry
> Thad on LinkedIn