[sc34wg3] Question on TNC / Montreal minutes

Steve Pepper sc34wg3@isotopicmaps.org
Mon, 16 Sep 2002 09:21:45 +0200

At 15:57 13/09/02 -0400, Patrick Durusau wrote:
>Steve Pepper wrote:
>>The basic problem is that sometimes you want to establish "contextual 
>>validity" without claiming that a name is unique within a certain 
>>namespace. Take the example of scoping by natural language:
>>   [table = "table" /english
>>          = "bord" /norwegian ]
>>All we want to do here is provide appropriate labels in English and 
>>Norwegian for those things with four legs. We don't want to claim that 
>>"table" is a name that can only refer to one subject in English (what 
>>about those things used to display data?), or that "bord" can only refer 
>>to one subject in Norwegian (it also means "plank").
>Hmmm, if I understand the example, you want to place "table" and "bord" in 
>different language scopes for something called "table"? So, "table" (first 
>occurrence in your example) is not claimed to be unique within some 
>larger, possibly unspecified namespace? In a sense a "mini-scope" around 
>the term table that limits the application of the language scopes to that term?

Sorry, I thought you were familiar with the LTM notation used in the 
example. It declares a topic (ID  "table") with two base names, in the 
scopes "english" and "norwegian", respectively. The use of scoped base 
names here is simply to allow applications to decide which name is most 
relevant, given a context in which you might have both English-speaking and 
Norwegian-speaking users. In this example, it is *not* the intention of the 
topic map author to establish "table" as a unique name within the namespace 
"English", but that is, in fact, what happens.

>If I am reading the example correctly, I am still missing the....Ah, are 
>you saying that the '"table /english' must apply only within the scope 
>(sorry no other word comes to mind) of language specific naming of the 
>term table and not to the term table in the larger topic map? Such that at 
>present, we have confusion between the operation of scope on '"table" 
>/english' and a scope applied to the term table? Interesting. Is that a 
>problem of syntax or understanding of the operation of scope? (Or both?)

'Yes' to the first part of this paragraph; 'not sure' to the rest.

We might have another topic, whose subject is "table (the device used to 
display data)":

   [table2 = "table" /english
           = "tabell" /norwegian ]

Again, we are just using scope to allow applications to choose the most 
pertinent name for the topic in the given language context, not to enable 
applications to unambiguously look up a single topic within some name 
space. If I'd wanted to do the latter, I would have had to use different 
scopes, e.g.:

   [table  = "table" /furniture ]
   [table2 = "table" /math ]

(Note: The scopes "furniture" and "math" were chosen without a lot of 
deliberation. I could have used any number of other scopes for the same 
purpose, perhaps "carpentry" and "information". The very arbitrariness of 
such choices is also one of the problems with the TNC and why it doesn't 
achieve the goal that its adherents have set for it, namely to permit the 
unambiguous look up of topics by name.)


Steve Pepper, Chief Executive Officer <pepper@ontopia.net>
Convenor, ISO/IEC JTC1/SC34/WG3  Editor, XTM (XML Topic Maps)
Ontopia AS, Waldemar Thranes gt. 98, N-0175 Oslo, Norway.
http://www.ontopia.net/ phone: +47-23233080 GSM: +47-90827246