Re: Temporal validitity of subject indicators?

Hi Andrew,

Am 07.05.2011 um 20:22 schrieb Andrew S. Townley:

> Hi folks,
> 
> Quick question arising from something I'm doing at the moment:  how do you invalidate/deprecate a subject indicator reference within a Topic Map ecosystem?

As far as i know the standards there is no such mechanism. A URL that is used as a subject identifier, identifies that subject once and for all.

> The issue here is that the URL used for a subject indicator can be updated over time to actually refer to a totally different subject--completely at the discretion of the owner.

There is two scenarios here. Person P1 equips a topic T which represents a subject S with an identifier I. Either

a. P1 has control over I or
b. P1 has not control over I, but rather a person P3

ad a: Control: Create a HTML page P (subject indicator) that when I is entered in a browser by a person P2, P2 will know that the page is about S for whichever reason. This works only when P1 and P2 have a strong shared context since the message otherwise will not be understood. E.g. they need to speak a common language which is used in P. For me it is in the very definition of a document that a it connects two subjects, so there needs to be a convention to recognize what is *the* subject. All these bits to the puzzle must be communicated out of band or be present in the shared context of culture.

ad b. No control: Then there is two cases: Either P1 trusts P3 to keep it constant and alive or not. In the latter, it is better not to use I as a subject identifier.

> What's the normal way to deal with this sort of thing?

The topic map ecosystem (interesting term which would need a definition, but i think i know what you mean :) can be intentionally or unintentionally polluted by misunderstandings, which are created when people think their subject is the one that is described in a subject indicator, but somebody else thinks it is a different subject. In a subsequent merging operations two statements about different subjects will land at the same topic. Bad.

The only way that i think this issue can be solved (in theory) is this: every topic map author must use his own subject identifiers since only then one can be sure that the statements he makes are at the right topic. He should only use exactly *one* subject identifier. I would call such topic maps, that hold information and use exactly one genuine subject identifier per topic, content topic maps. These are not dangerous, in the sense that they *cannot* contain misunderstandings.

Then you need some glue to connect it all. This glue is made up by glue topic maps, which:

a. hold only topics
b. every topic must have at least two subject identifiers

These topic maps must be authored with great care. But even the greatest care does not prevent misunderstandings. But this way the misunderstanding can be revoked by deleting it and rebooting the ecosystem.

Best,

Robert