|
> Date: Thu, 24 Jun 2004 17:52:44 +0200
> From: Theo van Veen <[log in to unmask]>
>
>>>> And ... why do you have kb.title and kb.author anyway? Surely
>>>> these are just dc.title and dc.author using 'exact' rather than a
>>>> word relation?
>>>
>>> Author is not in DC so nees to be in a separate indexset.
>>
>> No, the DC context-set does have an author index. It's called
>> "creator".
>
> As a legacy we have a different creator index and author index for
> different records in the same database.
Oh, OK. Then I think your decision to use a separate, private,
"author" index makes perfect sense.
> In the future however we might make a distinction between creators
> and their role. So when you search for author =xyz you will only
> find athors with that name but when you search for creator=xyz, you
> will find painters with that name, authors, composers and so on. As
> these roles are refinements of the dc:creator element I assume this
> index set should become part of a dcterms index set.
I'm not sure what dcterms is, but yes, somehow we should have a
standard way of expressing Qualified Dublin Core access points in
CQL. That could be done either with a set that defines a huge number
of indexes, or by defining mutant relation modifiers that act as what
the attribute architecture calls Semantic Qualifiers and Functional
Qualifiers. (I plan to think about this another day.)
> Actually we do not use the index prefixes, they are stripped
> immediately from the index names, but to define them in explain they
> need to be part of a set with a name.
Yup. It is good ZeeRex citizenship to hide your dirty laundry this
way :-)
_/|_ _______________________________________________________________
/o ) \/ Mike Taylor <[log in to unmask]> http://www.miketaylor.org.uk
)_v__/\ "Nyeeeoooowng Thhhhhhrb Ptooosh Ptoosh Wibba Wubba
Kaaaafooooooom" -- Harvey Thompson, imitating a Defender
machine.
--
Listen to free demos of soundtrack music for film, TV and radio
http://www.pipedreaming.org.uk/soundtrack/
|
