Fwd: Re: [topicmapmail] Topic map design guidelines?

Teresa Silio tsilio@cnic.es
Tue, 20 Jul 2004 11:15:51 +0200


This is a multi-part message in MIME format.

--bound1090314951
Content-Type: text/plain; charset=iso-8859-1
Content-Transfer-Encoding: 7bit

Hi all

Is it possible to set a set of primitive roles (in order to avoid variations) based on the Linguistic semantic roles?, that is:

Agent, Patient, Objetc, Source, Intrument, etc.

Thank you,
Teresa 

 Teresa Silio
 Head of Library and Information Science Services
 CNIC Fundacion Centro Nacional de Investigaciones Carlos III
 Ronda de Poniente, 5
 28760 Tres Cantos, Madrid
 Tel +34 91 8061880 x 1150
 Fax +34 91 8035258
 tsilio@cnic.es
 http://www.cnic.es


Jan Algermissen <jalgermissen@topicmapping.com> wrote ..
> schoenfeld.i@atlas.de wrote:
> > 
> > Hi all,
> > 
> > does anyone knows about a kind of style guide, guidelines or best practices
> of how to design good topic maps? 
> 
> Ingo--
> 
> I suggest that you look at some of the extensive material on data modeling,
> the general principles apply to data modeling with topic maps also.
> 
> Some stuff to consider:
> 
> - what are your classes? (although not a structural part of
>   the Topic Maps paradigm, classes are still a major design
>   element)
> 
> - don't mix classes and roles. 'Person' is a class, while 'Teacher'
>   is a role.
>   (see: http://www.infoloom.com/pipermail/topicmapmail/2002q1/003566.html
> )
> 
> - define the controlled vocabulary of your domain, avoid ambigous
>   names
> 
> - what are properties, what are relationships? Relationships
>   map directly to associations and participating roles provide
>   a fine secondary classification of your instances (e.g. "what
>   persons are teachers?" ==> find all players of role teacher)
> 
>   Properties of your instances (e.g. Age, Height, Address,...)
>   can only be expressed by tweaking the semantics of <occurrence>
>   a bit (in the absence of a dedicated property element in XTM):
> 
>   <topic>
>     <occurrence>
>       <instanceOf>
>         <subjectIndicatorRef xlink:href="http://.../Age" />
>       </instanceOf>
>       <resourceData>12</resourceData>
>     </occurrence>
>   </topic>
> 
>   This arguably does not leave much room for data typing the '12',
>   but there is currently no other way.
> 
>   Hmm...., you could combine your XTM with RDF, given the TM engine you
>   use supports this. E.g. along the following lines:
> 
>   In your XTM put:
>   <topic>
>     <subjectIdentity>
>       <subjectIndicatorRef xlink:href="http://people.org/mary" />
>     </subjectIdentity>
>     <!-- ... -->
>   </topic>
> 
>   In your RDF put:
> 
>   <rdf:description about="http://people.org/mary">
>      <foo:age>12</foo:age>
>   </rdf:description>
> 
>   Via RDF schema the semantics and data type of foo:age can be defined.
> 
> 
> - You should also (from my experience) restrict yourself to
> 
>   - using a given role only once in an association (no multiple
>     role players)
> 
>   - using a given role only in associations of a single type (the
>     combination of the roles is essentially what constitutes the
>     type of relationship, so they should not be mixed across
>     different types). This also enables you to select associations
>     soly by role: "find all assocications where X plays superclass"
>     as opposed to "find all assocications of type superclass-subclass
>     where X plays superclass".
> 
> Anyhow, just some guidelines I follow. Hope they help a bit.
> 
> 
> Jan
> 
>   
>     
> 
> Jan Algermissen                           http://www.topicmapping.com
> Consultant & Programmer	                  http://www.gooseworks.org
> _______________________________________________
> topicmapmail mailing list
> topicmapmail@infoloom.com
> http://www.infoloom.com/mailman/listinfo/topicmapmail

--bound1090314951--