Re: SKOS Guide and owl:sameAs

Hi Andrew!

>> Thanks for the examples and a smart solution with 
>> dct:replaces and dct:isReplacedBy. But this is just editorial 
>> information and does not directly help you in retrieval.
> 
> This type of information is more that editorial, it helps classifiers
> update the class numbers from one edition to another and can be useful
> during retrieval to find the appropriate class number in the latest
> or an earlier edition.

Maybe that's right - I thought versioning to be just a kind of mapping
but there should be some special treatment compared to mapping between
different Concept Schemes. I was just wondering because I have not seen
the use of dct:replaces and dct:isReplacedby with SKOS before.

>> I wonder what the relation to the SKOS mapping vocabulary should be.
>> In the first case:
>>
>> Edition 21
>>   skos:Concept 005.6
>>      mapping:exactMatch E22 005.18
>>
>> Edition 22
>>   skos:Concept 005.6
>>      mapping:exactMatch E22 005.18
> 
> No real issues here, but ...

So do you agree that dct:replaces is not meant to replace
skos-mapping-relations but there always needs to be a mapping (which you
may automatically derive from dct:replaces)?

>> In the second case:
>>
>> Edition 21
>>   skos:Concept T2--145
>>     mapping:narrowMatch E22 T2--145
>>     mapping:narrowMatch E22 T2--153
>>
>> Edition 22
>>   skos:Concept T2--145
>>     mapping:broadMatch E21 T2--145
>>   skos:Concept T2--153
>>     mapping:broadMatch E21 T2--145
> 
> Concept splitting is more problematic since it can be one to many,
> many to one, or many to many.  The problem above is that there is
> no coordination between the two narrowMatch to imply a one to many
> or many to many relationship.  Also, I cannot add additional
> structure to the SKOS smap:* elements.  Those elements are simple
> statements of fact, e.g., this concept is narrower/broader to that
> concept.

This is way clean support of coordination is strongly needed in SKOS.

>> Anyway dct:replaces and dct:isReplacedBy looks like one of 
>> these pitfalls where people can guess the meaning by the name 
>> of the property while it is not clearly defined. If one 
>> concept is replaced by another you still don't know how the 
>> meaning has changed - it's just said that something is 
>> dropped and something new has been created instead.
> 
> Yes, but I could have provided whatever information that might
> have been necessary by adding additional structure:
> 
> Edition 21
>   skos:Concept T2--145
>     dct:isReplacedBy
>       rdf:Alt
>         rdf:li
>           rdf:Description E22 T2--145
>             additional structure...
>         rdf:li
>           rdf:Description E22 T2--153
>             additional structure...

Please keep on providing such examples and ideas, so we will get a kind
of "Quick guide how to encode versioning in Concept Schemes with SKOS".
I hope that Alistair will find time to get back to discussion - there
have been some good suggestions on this mailing list and his thesis
contains more than the actual SKOS standard - I'm a little bit  	
impatient to collaboratively update the standard to get things like
coordination, notations, versioning etc. fixed.

Greetings,
Jakob

Received on Thursday, 2 November 2006 08:55:33 UTC