xml-dev - RE: Soft Landing

["Bullard, Claude L (Len)" <clbullar@ingr.com>]

In systems such as public safety, there are typically 
Master Indexes for:

Names
Locations
Vehicles

and these enable the network.  The names data is 
particularly nasty to maintain and use.  For example, 
given the name Robert Smith, how many of them do 
you think there are and how do you figure out which 
one is the one you are about to imprison for fifty 
years?  The process of identification is very difficult, 
usually comes down to a guess by the operator, and 
in severe cases, comes down to a DNA test.  That is why, 
BTW, a national DNA database is in the works and it is 
not to improve medical diagnosis.

Then there is performance.

Each one of these indexes is cross-referenced with different 
relationships.  The problem of the semantic network 
approach is to reuse information and not build queries 
that have to search so far and so deep into child tables 
to create a report or fill a QBE form that one waits 
until the end of time.  It is much harder than it looks.

Semantic networks are constrained in utility by the 
top level domain classification.  Ultimately, one 
ends up creating lots of semantic networks and 
they overlap.   One problem is isolating the overlaps 
and indexing these to ensure that performance does 
not degrade.  Harder than it looks.  The overlap 
can also lead to pernicious or superstitious results.   

Expert systems don't scale well and that is one reason for 
not betting the web's future on them.   When it 
gets down to defining what is "socially useful" 
you will find that there are many many definitions, 
cultural, political, financial and so on.  It 
can quickly become a very bad brew of power, greed, 
and zealotry (power under religion).  Caveat vendor.

Len 
http://www.mp3.com/LenBullard

Ekam sat.h, Vipraah bahudhaa vadanti.
Daamyata. Datta. Dayadhvam.h


-----Original Message-----
From: David Megginson [mailto:david@megginson.com]
Sent: Friday, October 20, 2000 5:29 PM
To: xml-dev@lists.xml.org
Subject: re: Soft Landing


Dave Winer writes:

 > I wonder if anyone is interested in trying to set up a
 > mini-Semantic Web of content describing the people on this list,
 > what their interests are, what software they use, who else they
 > know, etc. My poor little mind needs to try something pragmatic to
 > figure out what all this stuff means. Dave

Step one -- whether you're using RDF, Topic Maps, SOAP's serialization
format, or XMI, is to decide what your classes are and what your
properties are (note that I didn't say the classes and *their*
properties).  One of the classes is easy enough:

- Person (or to include orgs, etc., Party)

But what after that?  Should we have an Interest class to represent
shared interests, a WebSite class to represent Web sites, etc., or
should we represent them simply as text strings?  Here's the
difference, first with just text strings (denormalized):

  <xd:Person rdf:about="http://megginson.com/ids/p001">
   <xd:name>Dave Winer</xd:name>
   <xd:interest>XML</xd:interest>
   <xd:mailing-list>xml-dev</xd:mailing-list>
  </xd:Person>

  <xd:Person rdf:about="http://megginson.com/ids/p002">
   <xd:name>David Megginson</xd:name>
   <xd:interest>XML</xd:interest>
   <xd:mailing-list>xml-dev</xd:mailing-list>
  </xd:Person>

and then with more classes and objects (normalized):

  <xd:Person rdf:about="http://megginson.com/ids/p001">
   <xd:name>Dave Winer</xd:name>
   <xd:interest rdf:resource="http://megginson.com/ids/i001"/>
   <xd:mailing-list rdf:resource="http://megginson.com/ids/m001"/>
  </xd:Person>

  <xd:Person rdf:about="http://megginson.com/ids/p002">
   <xd:name>David Megginson</xd:name>
   <xd:interest rdf:resource="http://megginson.com/ids/i001"/>
   <xd:mailing-list rdf:resource="http://megginson.com/ids/m001"/>
  </xd:Person>

  <xd:Interest rdf:about="http://megginson.com/ids/i001">
   <xd:name>XML</xd:name>
  </xd:Interest>

  <xd:MailingList rdf:about="http://megginson.com/ids/m001">
   <xd:name>xml-dev</xd:name>
  </xd:MailingList>

The second one is obviously more flexible, but at the cost of
increased complexity -- in a real-world model, this would explode into
hundreds or thousands of objects for only a few people and their
interests.

Anyway, this is an example of the kinds of decisions people need to
make with or without a semantic web -- the design problem is where
implementors spend most of their time and money, and RDF or XMI or
Topic Maps or SOAP don't make that part any simpler.


All the best,


David

-- 
David Megginson                 david@megginson.com
           http://www.megginson.com/