Re: modeling location-of or birth-place-of

Thanks for this, Rob, it’s a big help. I think Option 1 is our answer – my constraint is that for this application (WHGazetteer) we need compatibility with Pelagios/Peripleo, which is OA, potentially indexing each other’s contributed annotations (TBD). I think we will favor JSON-LD, and interpret a few “standard” properties, maybe per-target_type. If contributions don’t include them, our interfaces will have to manage that somehow. E.g. a person could have an asserted connection to a place, but for an unknown interval. I like to avoid what I call the “dumb edge problem” but need to accommodate the realities of historical data.

 

Karl

 

On 1/1/18, 2:05 PM, "Robert Sanderson" <azaroth42@gmail.com> wrote:

 

 

Hi Karl,

 

There are three possible patterns for this, but first are Web Annotations really the right tool to be using for this?  It sounds like you want something with a stronger data model, rather than a framework for associating resources together.

 

Option 1:  Machine Readable Body

 

My preferred solution for these sorts of things is simply to have all of the assertions in the body of the annotation.  In the same way that you might write the place and time in English as a body, it's quite possible to write in XML, JSON, RDF or whatever other syntax is convenient.  This keeps the annotation assertions and the content assertions separate, allowing different agents to be responsible (and hence potentially credited) for them.  The downside is that you need to separately parse the body to understand the content [which is by design, but might be unpalatable].

 

Option 2: Create a new Motivation

 

Another option is to create a new motivation that mirrors the predicate/relationship you want to assert, such as yyy:asserting-place-of-birth or zzz:asserting-place-of-creation.

Then the place is the body and the thing receiving the place is the target. The advantage is that everything is parsed by your annotation system, the disadvantage is that now you've mirrored your entire ontology into a set of motivation extensions -- probably as narrower terms than oa:tagging.

 

Option 3: Create a new Property

 

Finally, you could simply add new properties to the model to describe the relationships you want to assert. The advantage is that you're not duplicating your ontology, the downside is that no other system will process your extensions.  The reason we didn't put this into the model is that RDF already has a reification pattern, which is generally unloved and we didn't want to bring it in to scope of the work, as we would start to compete with other standards like LDP.

 

Hope that helps,

 

Rob

 

 

 

 

 

 

 

 

On Sat, Dec 30, 2017 at 3:42 PM, karlg <karl.geog@gmail.com> wrote:

Hi,

 

I’m trying to model the annotation of a resource for any sort of thing with a resource for a place, and it needs to allow for a valid time. Can’t see how so far, any ideas would be helpful.

 

For example, annotating a record for a person (target) with a record for a place (body). The person wasn’t always at the place, so the relation between them needs to include a valid date range somehow. Maybe restricting the relation to type, e.g. “birthplace,” or “created-at” in the case of a creative work.

 

The goal is descriptions of places that include all sorts of items that have been annotated as being located there, generally, but critically also for some date range.

 

thanks

 

-- 

Karl Grossner, PhD

Technical Director, World-Historical Gazetteer

University of Pittsburgh World History Center

e: karl.geog@gmail.com

t: @kgeographer

 



 

-- 

Rob Sanderson

Semantic Architect

The Getty Trust

Los Angeles, CA 90049