[sc34wg3] Re: [topicmapmail] Multiple scopes on associations

[Jan Algermissen]

> [Lars Marius Garshol:]
> The issue is what happens when
> 
>   <topicMap>
> 
>     [...]
> 
>     <association>
>       <instanceOf>
>          <topicRef xlink:href="#A"/>
>       </instanceOf>
>       <scope>
>          <topicRef xlink:href="#THEME1"/>
>       </scope>
> 
>       <member>
>         <roleSpec>
>            <topicRef xlink:href="#B"/>
>         </roleSpec>
>         <topicRef xlink:href="#C"/>
>       </member>
> 
>       <member>
>         <roleSpec>
>            <topicRef xlink:href="#D"/>
>         </roleSpec>
>         <topicRef xlink:href="#E"/>
>       </member>
>     </association>
> 
>     <association>
>       <instanceOf>
>          <topicRef xlink:href="#A"/>
>       </instanceOf>
>       <scope>
>          <topicRef xlink:href="#THEME2"/>
>       </scope>
> 
>       <member>
>         <roleSpec>
>            <topicRef xlink:href="#B"/>
>         </roleSpec>
>         <topicRef xlink:href="#C"/>
>       </member>
> 
>       <member>
>         <roleSpec>
>            <topicRef xlink:href="#D"/>
>         </roleSpec>
>         <topicRef xlink:href="#E"/>
>       </member>
>     </association>
>   </topicMap>
> 
> is read in by a topic map processor. Do we get one association with
> two scopes, or two associations with one scope each? 13250, XTM 1.0,
> the infoset model, current topic map engines all think the latter,
> PMTM4 thinks the former.

IMHO the *real* question behind this issue is, whether two associations
that have equal roles/members are expressing the same subject or not.

I don't think that we can make the assumption that they are *allways*
expressing the same subject:

In Steven Necombs example Mary and John were married from 1983 until 1995 (B)
and they were married in? the religion "Roman Catholic Church" (A) [1]

It looks very reasonable to state that A and B are expressing the
same subject (the state of John and Mary being married to each other)
and that they should be merged on the basis of subject equality.

But what if John and Mary decide that getting divorced in '95 was wrong
and marry again in 2000. Then there would be association C of John and
Mary being married in the scope 2000-today. We now have TWO subjects:
their first marriage and their second and no automated application could
determine that from the roles and members in the associations.

My point here is, that the question whether two associations with equal
roles/members are expressing the same subject cannot be solved in all
situations without human intervention.

BUT (with a lot of emphasize) there will be (many?) associations in topic maps
that actually do express the same subject and it is likely that TM-authors will
demand the possibility to have single addressing points for these subjects
to reason about them. Well then, isn't it likely in a competitive world that
all the existing TM applications will be 'upgraded' sooner or later anyway in
order to support this important feature ?
(Take this -please- with a sense of humor!)


Jan

[1] This is my understanding of association A In an earlier posting
Martin Bryan understood A as being an event (The marriage-event between
John and Mary **IN** a Roman Catholic Church) I don't know which was the
original intention.