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

RE: Does data represent things besides entities, attributes, a

• From: "Michael Kay" <mike@saxonica.com>
• To: "'Costello, Roger L.'" <costello@mitre.org>,<xml-dev@l...>
• Date: Sat, 29 Aug 2009 15:57:12 +0100

> 
> > Data represents entities, attributes, and relationships.
>  
> Can every piece of data be categorized as a representation of 
> an entity, attribute, or relationship?
> 
> For example, here's a description of a book:
> 
>     In this groundbreaking book, evolutionary
>     biologist Jared Diamond stunningly dismantles
>     racially biased theories of human history by
>     revealing the environmental factors actually
>     responsible for history's broadcast patterns.
> 
> It is data. 
> 
> It doesn't seem to be a representation of either an entity, 
> attribute, or relationship.
> 

Actually, I think the tradition of formal data semantics, starting with
Abrial in 1974 and extending to RDF more recently, is that everything can be
reduced to entities and binary relationships between entities. An attribute
is merely a special case of a relationship where one of the entities is
something like a number or a character string. And binary relationships are
themselves entities, which can have their own attributes and relationships.

Where you are dealing with text like this, it's better to talk about
assertions rather than facts, and they often include qualifiers which are
assertions about the assertions. In particular, the above sentence is on at
least three levels: it's describing a theory ("environmental factors are
responsible for history's patterns"), it's asserting something that happened
to this theory ("Jared dismantles theory"), and it's saying where you can
find this happening ("Book contains Jared's dismantling of theory").

If you apply this to your sentence above, you get something like this:

Entity E1
E1 is-a book
E1 is groundbreaking
Entity E2
E2 is-a evolutionary biologist
E2 is-named Jared Diamond
Fact F0: E2 is-author-of E1
F0 is-a conjecture
Entity E3
E3 is-a-set-of theories
every E in E3 is a theory of human history
every E in E3 is racially biased
Fact F1: E2 dismantles E3
F1 is-done stunningly
Entity E4
E4 is a set of environmental factors
Entity E5
E5 is history's broadcast patterns (whatever that might be!)
E4 is responsible for E5
Fact F2: E2 reveals E4
F1 is-done-by-doing F2
F1 is-asserted-in E1

To be honest, I'm not sure how either Abrial or RDF handle the universal
quantifiers, but hopefully you get the idea. Basically, you're reducing the
sentence to a set of basic assertions and reducing it to predicate logic in
a number of variables, where the variables are the entities and the
predicates are the binary relationships.

Regards,

Michael Kay
http://www.saxonica.com/
http://twitter.com/michaelhkay 

• Follow-Ups:
  • Re: Does data represent things besides entities, attributes,and relationships?
    • From: Rick Jelliffe <rjelliffe@allette.com.au>

• References:
  • Does data represent things besides entities, attributes, andrelationships?
    • From: "Costello, Roger L." <costello@mitre.org>