OASIS Mailing List ArchivesView the OASIS mailing list archive below
or browse/search using MarkMail.

# xdi message

[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]

Subject: RE: Statement 6 is feeling weird..

• From: "Drummond Reed" <drummond.reed@cordance.net>
• To: "'Markus Sabadello'" <markus.sabadello@xdi.org>
• Date: Fri, 19 Dec 2008 21:40:41 -0800

```Funny you should say that, Markus. I was on a plane all day today flying
back from Boston to Seattle (left just before a big storm hit Boston; landed
in Chicago just after a big storm passed by; and reached Seattle just before
a big storm hits here - whew!)

Anyway, on the plane I went over Statement 6 again because it was bugging me
too, for exactly the reasons you identify. The solution was to revise the
graph for compound \$has relationships so that they chain directly. For
example, saying +x/\$has/+y and +y/\$has/+z means you can say:

+x+y/\$has/+z
+x/\$has/+y+z
+x+y+z

I have uploaded the new graphs that follow this logic. Links:

PDF

http://www.oasis-open.org/committees/download.php/30457/xdi-rdf-graphing-v2.
pdf
PPT

http://www.oasis-open.org/committees/download.php/30456/xdi-rdf-graphing-v2.
ppt

NOTE THAT THE ORDER OF THE DIAGRAMS CHANGED - with the new logic, it made
much more sense to move #6 in V1 to become #4 in V2.

Besides solving the issue you identified, this logic has two other
advantages:

1) It keeps the rules 100% clean: \$has relationships between two XRI can
ALWAYS be expressed by direct concatention of the subject and the object,
with NO CHANGE to either.

2) It reflects the same property in natural language (at least in English
anyway). For example, if I use the following four nouns as a phrase...

American company president signature

...then it still means the same thing if I parse them into any of the
following combinations:

American + (company president signature)
(American company) + (president signature)
(American company president) + signature

So this way our logic lines up with natural language too.

MUCH better. Do you agree?

=Drummond

> -----Original Message-----
> From: markus.sabadello@gmail.com [mailto:markus.sabadello@gmail.com]
> On Behalf Of Markus Sabadello
> Sent: Friday, December 19, 2008 6:24 PM
> To: Drummond Reed
> Cc: OASIS - XDI TC
> Subject: Statement 6 is feeling weird..
>
> In: http://www.oasis-open.org/committees/download.php/30442/xdi-rdf-
> graphing-v1.pdf
>
> First of all, the rules for \$has and \$has\$a used to be quite simple..
> Now with Statement 6, they become a bit harder to explain.
>
> In addition, it also feels weird on a technical level. There's
> something that bugs me. Unfortunately I'm not a
> mathematician/linguist/etc, but let me put it into two examples and a
> few questions:
>
> ---
>
> Example 1
>
> What statement produces the XRI +x(+y) ?
>
> Is it 1A) +x/\$has/+y, or is it 1B) +x/\$has/(+y) ?
>
> It can't really be 1A), because the XRI produced from 1A) is +x+y,
> according to Statement 4 in the doc. So it must be 1B).
>
> ---
>
> Example 2
>
> What XRI is produced by the statement +x/\$has/(+y+z)?
>
> Is it 2A) +x(+y+z), or is it 2B) +x((+y+z)) ?
>
> 2A) feels weird, because that XRI is already produced by +x/\$has/+y+z,
> according to Statement 6 in the doc.
> 2B) feels weird, because why would you have to put (+y+z) in another
> set of parens, if they already are in their own context.
>
> Note that the statement in Example 2 is similar to the statement 1B)
> in Example 1. In both cases the object is just a single subsegment.
>
> ---
>
> Does that make any sense? If you still think Statement 6 is correct,
> then maybe you could add these 2 examples to your Graphing doc?
>
> Markus

```

[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]