IVOAT Top Level Concepts

[Alasdair Gray]

Alasdair Gray wrote:
> Hi Rick,
>
>   
>> Nominally, the script notes tokens which are NT's of other tokens,  
>> indicating that they are not top-level, and doesn't spit these out in
>>     
>
>   
>> the TopConcept list. 
>>     
>
> From this explanation, I must conclude that there is an error somewhere
> then. Consider IVOAT:cataclysmicvariablestars. It appears both as a
> narrower term of IVOAT:variablestars and as a topLevelConcept.
>
>   
>> If we can produce a version of the text list  
>> (e.g. via application of cwm) where the NT's and BT's are correctly  
>> noted (i.e. if X  BT Y, then Y  NT X) then the list will at least be  
>> self-consistent.  Someone already mentioned that there is lots of  
>> room to reduce the number of top concepts dramatically, but this  
>> cannot be done by a script.....
>>     
>
> We'll need to do some more work on this.
>   
I have spent some time looking at the original IAU Thesaurus, 
considering the top level concepts. If we follow Rick's suggestion that 
a top level concept is one that does not appear as a narrower term in 
any other concept then we will end up with a different poly-hierarchy 
than the original IAU Thesaurus. Let me illustrate what I mean with an 
example.

In the original hierarchy definition for the IAU Thesaurus you will find 
the following:
A STARS
.     A DWARF STARS
.     A GIANT STARS
.     A SUBDWARF STARS
.     A SUBGIANT STARS
.     A SUPERGIANT STARS
.     Ae STARS
.     Am STARS
.     .     SPECTRUM VARIABLE STARS
.     .     .     MANGANESE STARS
.     Ap STARS
.     .     HOLMIUM STARS
.     .     MANGANESE STARS
.     .     SILICON STARS
.     .     SPECTRUM VARIABLE STARS
.     .     .     MANGANESE STARS

[snip]

EARLY TYPE STARS
.     A STARS
.     .     A DWARF STARS
.     .     A GIANT STARS
.     .     A SUBDWARF STARS
.     .     A SUBGIANT STARS
.     .     A SUPERGIANT STARS
.     .     Ae STARS
.     .     Am STARS
.     .     .     SPECTRUM VARIABLE STARS
.     .     .     .     MANGANESE STARS
.     .     Ap STARS
.     .     .     HOLMIUM STARS
.     .     .     MANGANESE STARS
.     .     .     SILICON STARS
.     .     .     SPECTRUM VARIABLE STARS
.     .     .     .     MANGANESE STARS
.     B STARS
[snip]
ULTRAVIOLET SOURCES
.     EARLY TYPE STARS
.     .     A STARS
.     .     .     A DWARF STARS
.     .     .     A GIANT STARS
.     .     .     A SUBDWARF STARS
.     .     .     A SUBGIANT STARS
.     .     .     A SUPERGIANT STARS
.     .     .     Ae STARS
.     .     .     Am STARS
.     .     .     .     SPECTRUM VARIABLE STARS
.     .     .     .     .     MANGANESE STARS
.     .     .     Ap STARS
.     .     .     .     HOLMIUM STARS
.     .     .     .     MANGANESE STARS
.     .     .     .     SILICON STARS
.     .     .     .     SPECTRUM VARIABLE STARS
.     .     .     .     .     MANGANESE STARS
.     .     B STARS
[snip]

My understanding of this is that A STARS is a top level concept (from 
the first declaration) that also appears as narrower term than EARLY 
TYPE STARS (from the second declaration) which in turn is both a top 
level concept and a narrower term than ULTRAVIOLET SOURCE (from the 
second and third declarations).

 From Rick's definition, only ULTRAVIOLET SOURCE would appear as a top 
level concept.

I personally find the original hierarchy unsatisfying due to the large 
number of top level concepts which makes it hard to navigate and 
understand the relationships between the concepts. I like Rick's 
definition that a top level concept is one that is not a narrower term 
than another concept. It matches my intuition of what should happen.

 From a SKOS point of view (according to the current core guide) the 
following is stated:

    For concept schemes where the concepts are arranged in a
    broader/narrower generalisation hierarchy, use the
    skos:hasTopConcept property to assert a link between the concept
    scheme and the concepts that are the top-level concepts in the
    generalisation hierarchy.

    It is recommended that you use the |skos:hasTopConcept
    <http://www.w3.org/2004/02/skos/core/spec/#hasTopConcept>| property,
    as this gives applications an efficient way of locating the top
    level concepts for a given scheme.

This means that it does not really matter which of the approaches we follow.

So, the purpose of this email was to highlight that we are deviating 
from the original IAU Thesaurus poly-hierarchy and to argue that this is 
a good thing and that we should keep going this way. Does anyone object?

Alasdair
>
> Alasdair J G Gray
> Research Associate: Explicator Project
> http://explicator.dcs.gla.ac.uk
> Computer Science, University of Glasgow
> 0141 330 6292
>   

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.ivoa.net/pipermail/semantics/attachments/20071102/7b486530/attachment-0001.html>