[XML-DEV Mailing List Archive Home] [By Thread] [By Date] [Recent Entries] [Reply To This Message]

RE: On edge-labelled graphs in XML

• From: Michael Rys <mrys@m...>
• To: 'Wayne Steele' <xmlmaster@h...>, xml-dev@l...
• Date: Thu, 07 Sep 2000 13:10:30 -0700

The main reason is that not everything is a tree. Information has the
tendency to organize itself in graphs of basically entities (nodes) and
their relationships (edges). Edge-labelled models assume that nodes have a
logical identity and that the name is part of their relationship. There may
be different kinds of relationship (i.e., containment, reference,
attributes, 1-1, 1-n, m-n etc.). 

The above is especially true if you talk about information items that are
arranged into different views. Even if the views are tree-views (like XML),
the overall structure of the information (the union of all views) is still a
graph. E.g., data that describes me may have data about my mother as the
parent in the tree (relationship child) in one view, and my bosses data as
parent in another. If you want to express these relationships in
hierarchies, you can for each view, but the overall information organization
is a graph.

Best regards
Michael

PS: I left out that I favor the edge-based abstraction for the query
datamodel of course in my message cited below...

> -----Original Message-----
> From: Wayne Steele [mailto:xmlmaster@h...]
> Sent: Thursday, September 07, 2000 12:47 PM
> To: Michael Rys; xml-dev@l...
> Subject: RE: On edge-labelled graphs in XML
> 
> 
> Why would someone prefer an edge-based model to a node-based one?
> or vice-versa?
> 
> IMHO, the concepts of a "node" and a "tree" seem pretty 
> straightforward; 
> dealing with edge-based abstractions is about as comfortable 
> as reading a 
> book turned upside-down.
> 
> -Wayne Steele
> 
> 
> >From: Michael Rys <mrys@m...>
> >To: "'tpassin@h...'" <tpassin@h...>, xml-dev@x...
> >Subject: RE: Research notebook: On edge-labelled graphs in XML
> >Date: Wed, 06 Sep 2000 23:36:11 -0700
> >
> >This is an interesting observation. As a proponent of 
> edge-labelled graphs
> >(Lore, proposed XML data model based on edge-labelled graphs 
> about a year
> >ago) I would favor this, however, due to the extensive 
> node-centric models
> >within the W3C (DOM, XPath, but not the Infoset which should 
> be interpreted
> >as agnostic on this), the current working draft is (in my 
> personal and
> >humble opinion unfortunately) also node-labelled...
> >
> >Best regards
> >Michael
> >
> ______________________________________________________________
> ___________
> Get Your Private, Free E-mail from MSN Hotmail at 
http://www.hotmail.com.

Share information about yourself, create your own public profile at 
http://profiles.msn.com.

• Prev by Date: RE: UDDI, uddi.org and an open source project at SourceForge
• Next by Date: Re: On edge-labelled graphs in XML
• Previous by thread: Re: On edge-labelled graphs in XML
• Next by thread: Re: XML and LDAP: Common APIs?
• Index(es):
  • Date
  • Thread