On Jul 12, 2011, at 9:43 AM, glenn mcdonald wrote:
In common JSON usage, consider the following:
{
"@subject": "http://greggkellogg.net/foaf#me",
"@type": "Person"
"home-address": {
"@type": "Address",
"city": "San Rafael",
"region": "California"
}
}
Having to specify an IRI for the "home-address" item runs against much common usage. Why should this usage be discouraged?
This is fine data, but it's not a Linked Data graph. That is, it's not linked (can't refer to the address, city or region (or either type) as nodes), and it's a tree rather than a graph. Are we trying to magically transform JSON trees into graphs, or are we trying to describe how people with graphs should publish them in JSON? Or something else?
glenn
Sorry, I as assuming that Address and Person were terms that expand to IRIs. Values for city and region are typically literals in common formats. Actually, in schema.org<http://schema.org>, it's PostalAddress and addressLocality/addressRegion instead of city/region, but I was trying to be illustrative, not pedantic.
This short example is, of course, a tree, but could be a graph if I started adding more information, such as 'knows' with a couple of other people. Those people could either be described internally, or be their own documents which could express that they know me too, which would make it a graph.
The question is, do unnamed nodes have a place in Linked Data and in JSON-LD? I think they do; anyone else?
Gregg
Received on Tuesday, 12 July 2011 16:59:33 UTC