Re: Temporal validitity of subject indicators?

Robert,

OK, I think I understand what you are saying, but I am hard pressed to 
see how your "glue" topic maps avoid the problem.

At some point, some author of the "glue" topic maps has to choose a 
"local" subject identifier and a "foreign" identifier (one he did not 
author).

How is that author any more privileged than any ordinary author to avoid 
error?

I freely grant that what ensues is the "Tower of Babel" but then topic 
maps were never a solution to that problem.  They can only lessen its 
impact in particular contexts. Among people willing to commit the effort 
for the benefit of collating information from diverse sources.

To put it another way, there isn't ever a point a which we have it all 
"straightened out" and free from error. Semantics change on a daily 
basis and the most we can hope for is to make particular systems run 
with a little less semantic impedance. And those will require maintenance.

Hope you are having  a great weekend!

Patrick



On 5/8/2011 4:44 PM, Robert Cerny wrote:
> Patrick,
>
> Am 08.05.2011 um 20:22 schrieb Patrick Durusau:
>
>> Robert,
>>
>> On 5/8/2011 12:49 PM, Robert Cerny wrote:
>>> Patrick,
>>>
>>> Am 08.05.2011 um 17:42 schrieb Patrick Durusau:
>>>
>>>> I am intrigued by your:
>>>>> since you can always add a glue map, while it is impossible to get the mess apart.
>>>> Why so?
>>>>
>>>> Shouldn't I be about to "undo" a merge like I do editing?
>>> The mess is not caused by a merge operation, the mess is caused because a person reuses but misunderstands a (foreign) subject identifier and attaches statements to a topic which they actually do not belong to. Once both interpretations reach the Topic Maps ecosystem (great term, Andrew!), the damage is done. This you cannot undo no matter how smart you conceive your software, because the software is doing everything right. The human makes the error. Under very special circumstances a correction might be possible, but not in the general case.
>>>
>> Ah, sorry, I mis-read your complaint.
>>
>> What you mean to say is that we only have the grouping of subject identifiers and we lack a mechanism to *exclude* a subject identifier.
>>
>> Is that the problem?
> No, my problem is a generalization of Andrew's original question: what happens if a subject indicator is changed and indicates subject S1 at time T1 and subject S2 at subject T2. The general problem is that every time a person reuses a subject identifier SI and decides that the subject he has in mind is the same as the creator of SI had in mind, an error can happen. You cannot prevent this with any computational means.
>
> Then you need to put into account that we are talking about the Topic Maps ecosystem, which i would define as a TMDM engine where you throw in all topic maps that you discover and have them merged. All the misunderstandings will be collected there. You could not trust the information. It would be full of contradictions if you take a lot (>10000) of active authors who anarchically reuse subject identifiers. Even if none of them has bad intentions, it would be a tower of babel.
>
>>> Needless to say that these problems apply to any form of reusing foreign identifiers, so it also concerns the Linked Data technology. I am not sure if a similar solution to my separation in content and glue topic maps makes sense there. I do not know the involved technologies well enough.
>>>
>> I don't think it applies to Linked Data.
>>
>> Mostly because there is no guarantee that the links are about the same subject or the results of following one are meaningful.
> Neither is there a guarantee in Topic Maps that subject identifiers are about the same subject since different people may "interpret" it differently by accident. People make errors and only people assign subject identifiers to topics.
>
>> You sort it all out after gathering up the "linked data."
> You can't because the error already happens when a person misinterprets an URI. You cannot sort it out.
>
> I can give you a more elaborate example later this week, if i still could not make myself clear.
>
> Kind regards,
>
> Robert

-- 
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