Re: Nested Sets vs. Nested Intervals

From: <>
Date: 9 Nov 2005 14:54:49 -0800
Message-ID: <>

Semantic domain is as you know the fancy name for the "things" you represent with your schema and data.

Relational *key* people (sorry, "star" was bad choice of word, unless I cited Kleene too:-) you know and cited two, Codd, Chen, and then there's Date, Ulman, Wiederhold, etc. and I would not mind citing here the people siting at the W3C working groups for XQuery, XPath, etc.

E-R maps better to graph-based representations than to relational ones. It is not a matter of "can" but of "how well."

No I won't mention UML nor say the dirty word ("semi-structured":-)

There's lots of (good) graph-based models. RDF is (very) different from XML and both are (very) different from Codasyl and my baby called Mneson is (very, not so much, less) different from Codasyl, XML, RDF. (I suggest Mneson be discuss in the thread "Implementing a graph algebra" because that's it)

The "network" model in particular, or Codasyl: nodes are tables with records and attributes. Not a pure graph. Actually not very graph-like at all. I think the technical term is "a dirty mess."

How is RDF or XML different from Codasyl in particular: a lot different. An RDF node is a single concept, not a table like in Codasyl. An RDF edge is a single property out of a possibly (depending on the meta-schema) open set of properties, not the single relation "child" of Codasyl. Etc. RDF is a much cleaner model. A true graph model. This has good consequences at many levels (theory, practice, adoption). For example you can find very nice algebras for RDF out there. Received on Wed Nov 09 2005 - 23:54:49 CET

Original text of this message