ontolog-forum
[Top] [All Lists]

Re: [ontolog-forum] Data & Relations

To: "[ontolog-forum]" <ontolog-forum@xxxxxxxxxxxxxxxx>
From: Stephen Young <steve@xxxxxxxxxxxxxxxx>
Date: Thu, 30 May 2013 09:42:24 +1000
Message-id: <CAHH+T2+-25ufrOS81G4HKGAqcKL-4Vhks4_U_11hJw=5gpGNow@xxxxxxxxxxxxxx>
As a Graph Database vendor (GraphBase), this is topic close to my heart (and using Kingsley's many posts as a model, I'll try to keep the vendor-specific stuff to a minimum ;-) ).

In my experience, it's difficult enough working through the simple modelling problems associated with representing data-as-a-graph, without introducing pretty-much *any* of the formalisms that an ontologist takes for granted.

Futhermore, many of the RDBMS mapping problems we see have printed ER models the size of a physical desktop.  Few architects (in my opinion) will be willing to define the resulting structures using those formalisms.  Particularly given that the reason they're entertaining an RDBMS alternative in the first place is that they want some agility in how they structure their data.

The approach we've taken is to start with a base of representative problems and to "dumb down" the formalism and the structures needed to support it.  Unsurprisingly, they're all "data management" problems so the data structures don't need to be particularly expressive.

But even with these basic structures, RDBMS->Graph is difficult for most Enterprise technical people to get their head around.  We've had to provide abstractions that simplify working with graph-structured data even further - and those abstractions take them even further from the underlying formalisms.

And to insult the ontologist even further, we've had to build tools which suck data directly from an RDBMS and automatically create a data-graph - without any input from the architect.  Something that they can play with straight away.  This seems to be the best way to get engagement.


On 30 May 2013 04:30, Kingsley Idehen <kidehen@xxxxxxxxxxxxxx> wrote:
On 5/22/13 2:01 PM, Kingsley Idehen wrote:
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.

Yes, but mapping that out isn't so easy, when you have an attention challenged non technical audience as the target. For instance, many that even work with RDF wouldn't instinctively associate the IRI in the predicate role/slot of an RDF triple with relation name i.e., what denotes the relation represented by said RDF 3-tuple.

Likewise, deconstructing an RDBMS table to unveil it too is comprised of 3-tuples where:

1. Table Name denotes Domain
2. Field/Column Name denotes Predicate
3. Field/Column data types supported by DBMS determine Range.

John and others,

Here is are some vital characteristics upon which one could attempt to build a comparison (for moderately technical and attention challenged audience) between relational tables oriented RDBMS systems and relational (property/predicate) graph oriented RDBMS systems:

1. Relationship Representation
2. Relation Representation
3. Identifiers Types
4. Data Value Types (aka. Datatypes)
5. Entity Relationship Semantics Granularity.

Are there missing characteristics from the list above, bearing in mind the target audience?


--

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







_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J
 



--
Stephen Young



_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/  
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/  
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/ 
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J    (01)

<Prev in Thread] Current Thread [Next in Thread>