Re: Property Graphs as (bound) RDF Datasets

On 6/5/13 1:18 AM, Sandro Hawke wrote:
> Has anyone done a careful analysis of how Property Graphs [1] relate 
> to RDF?   They are getting an enormous amount of buzz at the moment 
> (eg here at SemTech).   I'd really like to be able to tell people 
> they're a subset of RDF Datasets, and I think they are if we adopt 
> something like the bound datasets proposal [2].
>
> Below, I've expressed in TriG the upper-left portion of the property 
> graph shown in [1].
>
> PREFIX : <>    # just make all the ids be relative, for now
>
> { # flag this as a bound dataset
> <> a <http://www.w3.org/1999/02/22-syntax-ns#BoundDataset>. }
>
> {  # vertex properties are datatype properties:
> :v1 :name "vadas";  :age 27.
> :v2 :name "marko"; :age 29.
> :v3 :name "lop";      :lang "java". }
>
> # edges go in named graphs
> # edge ids are converted to graph names
> # edge labels become object properties
> :e9 { :v1 :created :v3 }
> :e7 { :v1 :knows :v2 }
>
> {  # edge properties are datatype properties of the named graphs
> :e7 :weight 0.5.
> :e9 :weight 0.4 }
>
> This seems pretty clean to me, but perhaps I'm missing something. I 
> haven't actually done any Property Graph programming yet.
>
>        -- Sandro
>
>
> [1] https://github.com/tinkerpop/blueprints/wiki/Property-Graph-Model 
> (and attached in case in changes)
> [2] http://lists.w3.org/Archives/Public/public-rdf-wg/2013Jun/0020.html
>

Why would you be missing anything? Especially if the following statement 
is true:

"In a relational DB, for example, a relation is represented by a table 
and a relationship is one row of the table.  In RDF,
a relationship is a single triple, and a relation is the set of all the 
triples with the same relation name." [1].

RDF is supposed to be about the ability to discern meaning from 
*explicit* machine- and human-comprehensible entity relationship 
semantics. Thus, it shouldn't be victimized by fluffy marketing driven 
relabeling of its own core concepts :-)



Links:

1. http://ontolog.cim3.net/forum/ontolog-forum/2013-05/msg00196.html -- 
thread on Ontology about reconciling relational database terminology

-- 

Regards,

Kingsley Idehen	
Founder & CEO
OpenLink Software
Company Web: http://www.openlinksw.com
Personal Weblog: http://www.openlinksw.com/blog/~kidehen
Twitter/Identi.ca handle: @kidehen
Google+ Profile: https://plus.google.com/112399767740508618350/about
LinkedIn Profile: http://www.linkedin.com/in/kidehen