Re: Dynamic vs. Fixed World Views was Re: MARCXML to Topic Maps? MODS to Topic Maps?

Lars,

On 7/18/2010 7:06 AM, Lars Heuer wrote:
> Hi Patrick,
>
> [...]
>    
>> Reflecting my personal preferences as any "subject recognizer" reflects
>> its creator's preferences, it "chooses" to say this is an occurrence (in
>> the topic map sense) of the topic "Ex nihilo" that has the subject
>> identifier, http://en.wikipedia.org/wiki/Ex_nihilo.
>>      
> Well, I liked your pragmatic approach since I am a pragmatic person,
> but now I think Alexander made some disillusioning points. ;)
>
>    
Not really. Alex has mistaken a presentation for something of substance. 
The issue of recognizing subjects is the real issue.
> It is very true that it does not help to create just another container
> with mess. The mess wouldn't get any better if it is mapped into Topic
> Maps.
>
> Topic Maps excels in identifying subjects but if we have no chance to
> identify the subjects in MARC we're lost.
>
>    
True, but that is the case for *any* data set, not just MARC records.

If you can't identify the subjects in a dump of stock exchange data, 
census records, etc., a topic map isn't going to do you much good.
> A mapping to occurrences is a possibility, but I wonder if that has
> any value. (Someone (Kal?!?) said years ago that occurrences are the
> dead ends of Topic Maps. Very true).
>
> Take our discussion: A subject "Patrick" (identified by an e-mail
> address) is talking about subjects (MARC, Topic Maps) to other
> subjects (identified by e-mail addresses). That makes sense. But if we
> simply create topics from each post (with a random identifier) and put
> everything into occurrences, we achieved nothing. We can make wild
> guesses from the occurrences and NLP may help us a bit, but we
> wouldn't recognize the *subjects*.
>
>    
Err, somewhere in that paragraph I got lost.

If you recognize the subject "Patrick" then it would have a subject 
identifier that would trigger merging with subjects recognized from [ 
pdurusau-/[email protected], [email protected], [email protected], 
patrick-Q/[email protected]] so you have a "subject" that has associations with 
other subjects. Same happens for MARC and Topic Maps as well as any 
other subjects. The "occurrence" is the presence of a subject in a 
particular record, where the record is also a subject. (This is only one 
way model this and there are probably better ones, like treating the 
item identified by the record as the subject and the record as a 
separate subject.)

If you take the "view" approach to topic maps then you could 
interactively "recognize" subjects and see how that impacts the 
resulting topic map. Just a thought.

> If we put everything into occurrences, we could also use a full-text
> search on top of MARC and we'd have the same result. No need to create a
> Topic Maps silo at all.
>
>    

Maybe you need to explain what you mean by "put everything into 
occurrences." I may be missing what you mean by that.

I don't think of topic maps as silos and in my view they should not be, 
unless of course a paying customer wants a silo. I can't imagine the 
folks at Y-12 (they build atomic bombs) wanting to have anything other 
than a silo and good for them. ;-)

Hope you are at the start of a great week!

Patrick


> Best regards,
> Lars
>    

-- 
Patrick Durusau
patrick-Q/[email protected]
Chair, V1 - US TAG to JTC 1/SC 34
Convener, JTC 1/SC 34/WG 3 (Topic Maps)
Editor, OpenDocument Format TC (OASIS), Project Editor ISO/IEC 26300
Co-Editor, ISO/IEC 13250-1, 13250-5 (Topic Maps)

Another Word For It (blog): http://tm.durusau.net
Homepage: http://www.durusau.net
Twitter: patrickDurusau