[topicmapmail] Temporality in Topic Maps

[Simon Grant]

I'm not (yet :-) ) an expert on Topic Maps but I do have some of the 
same concerns.
I expect to be considering how TM can play a part in e-portfolio systems.
To me, it depends on the semantic role of temporality.

In an e-portfolio setting, we will need to deal with all kinds of 
records stored "out there" on various servers (who knows - blogs, 
social software, whatever). Because it seems unnecessary and perhaps 
undesirable to duplicate these records within a Topic Maps system, 
perhaps they are best seen as occurrences - resources - rather than 
topics. I don't know what sufficient reasons would be to motivate 
representing any occurrences as topics in their own right.

Perhaps one can distinguish two kinds of temporal "metadata" (though 
I use that word with great caution) or attribute. One is 
fundamentally inherent to what is recorded. In the case of a record 
of an action, for example, the start and finish times/dates are 
clearly to me inherent in the action. The other kind of attribute 
could be seen as a superstructure or overlay of personal 
significance. This would be of the type that a certain action (etc) 
is significant as evidence towards something I want to share - say my 
ability as a Topic Maps analyst. I see that as *not* inherent in the 
record itself, and ideal for representing with TM.

The confusing case - here is the semantic difference - would come if 
I were, for instance, a historian. I might have a particular interest 
in "The 17th Century" (or, equivalently, the years 1600 - 1700). This 
would be best represented as a topic, so that I can work it in to my 
structures of personal significance.

This brings me on to another angle. It may be of interest to me to 
represent all kinds of things as topics of personal interest, but 
alone, that remains a lonely situation. If my topics are either to be 
shared with others meaningfully, or (perhaps even more likely) to 
originate from a pool of common, shared meaning, then it would be 
futile to have any hour of any day of any year as a topic. The topics 
would not coincide enough to be of use or help or even interest. It's 
a little like the threshold for getting into Wikipedia.

In summary, I'm saying that my current view is that temporality will 
be a record-centric attribute in most situations that I can think of, 
much like location. But when a particular time period is the focus of 
common, or potentially shared, interest, it might well be worth 
representing a time period (or occasionally a moment, like the moment 
you heard that Princess Diana had died) as a topic.

I hope this makes some sense to you more experienced TMers.

Simon

At 03:08 2006-06-04, you wrote:
>Hi all,
>
>I've been struggeling lately with the concept of temporality in KM
>systems and was wondering what people here normally do when they face
>these issues, such as relationships, topics and reifications over
>time. Would you do one association with a role for time, or
>occurrences of the various topics, etc?
>
>topic 1990-1993
>topic 1993-1995
>topic 1995-1999
>topic 1999-2006
>
>assoc X
>   type membership
>   member "some person" role:human
>   member "some organisation" role:organisation
>   member 1990-1993 role:time-period
>   member 1999-2006 role:time-period
>
>This ends up hopelessly complex when you've got huge numbers. The other way ;
>
>topic "some person"
>   occurrence "membership-at-some-organisation" value="1990-1993"
>   occurrence "membership-at-some-organisation" value="1999-2006"
>
>Again, quite messy. Let's move on to ;
>
>topic "some person"
>topic "some organisation"
>topic "some period"
>assoc X
>   type membership-period
>   member "some person" role:person
>   member "some organisation" role:organisation
>   member "some time period" role:time-period
>
>I assume most would lean towards this latter, but this also presents
>problems with scalability and when "some organisation" changes over
>time, especially where there's overlap in "some period".
>
>How does people handle these things?
>
>
>Alex
>--
>"Ultimately, all things are known because you want to believe you know."
>                                                         - Frank Herbert
>__ http://shelter.nu/ __________________________________________________
>_______________________________________________
>topicmapmail mailing list
>topicmapmail at infoloom.com
>http://www.infoloom.com/mailman/listinfo/topicmapmail