Re: Temporal validitity of subject indicators?

Hi Lars,

Great that you join the discussion!

Am 10.05.2011 um 18:01 schrieb Lars Heuer:

> Hi Robert,
> 
> Unsure if this fits to Andrew's original question, but you mentioned
> "glue maps", so let me express my unqualified opinion. ;)
> 
> [...]
>> a. hold only topics
>> b. every topic must have at least two subject identifiers
> [...]
>> b. every topic must have at least two subject identifiers and no other characteristics
> 
> Glue topic maps just delay the problem. They are nice if you think
> "one topic map per doc"-centric and you have complete control over
> your environment and you can ensure that nobody writes an evil glue
> map which merges your complete topic map into one topic.
> 
> I think Patrick is right that "merging" is the problem, or the lack of
> unmerge.

No, not at all. That is not the problem that i am talking about. The problem i am talking about is of a semiotic nature and does not lie in the realm of technology or semantics. It applies to any reuse of identifiers, but in particular to the current case of reusing URL in linked data or Topic Maps.

The error that i am talking about happens when a person (the only entity who can assign subject identifiers) mistakenly uses subject identifier A to make statements about subject B. If the "identity distance" between subjects is close. They are hard to distinguish (which the topic map author might not be aware of).

Let me illustrate it with an example: topic map author A1 has a person P1 in mind. He discovers a subject identifier SI1 and thinks "Great, that is P1". But actually it is P2. He creates a topic map an uses SI1 but makes statements about P1. This "act of polluting the subject identifier SI1" in the Topic Maps ecosystem is not a problem of merging, it is a problem of reusing foreign identifiers.

> What's nice about RDF: They don't merge. :) You can still get a
> subject-centric view from RDF (collect all statements where X is the
> subject or object).

Let me repeat: it has nothing to do with the semantic technology. The same problem applies to RDF if identifiers are reused.

The problem i am talking about *can* be solved by the approach i sketched. That there is other subsequent problems which arise when aggregated statements are redistributed in the eco system is a different story.

Best,

Robert