Re: [TF:DbE] The easiest keys there are

On Sep 28, 2007, at 6:51 PM, Bijan Parsia wrote:

> Hi folks,
>
> The OWLED task force on DatabasEsque features:
> 	http://code.google.com/p/owl1-1/wiki/DatabasEsque
>
> Well, at least Uli and me, have been doing a bit of work on keys  
> (aka, inverseFunctional datatype properties) prompted by a visit to  
> Manchester by Matthew Pocock. Some sort of keys is a pretty high  
> value feature. However, if you check out this poster:
> 	http://webont.org/owled/taskforces/dbe/keys_poster.pdf
>
> (Bit of explanation: Bits in the tan clouds correspond to the  
> asserted parts of the kb. So you see the named entities, e.g., m1  
> or s1, in there. s1, for example, is known to have a key, but we  
> don't know what the key is.)
>
> ===SEMANTICS===
[snip]

Oops. Part of my message got omitted (as indicated by the uncompleted  
sentence). Sorry about that. The poster, as it stands, is too  
confusing to work through directly, IMHO. (E.g., I had trouble  
working through it :)). It was a poster designed in some haste for  
OWLED and to be support someone explaining various options. My point  
in posting a link to it is to help give a sense that keys, in  
general, are a non-trivial addition to the language.

The proposal I outlined is for a restricted form of first order key  
reasoning (i.e., no checking or working with missing keys, working on  
named individuals only, limited or no use of keys in class  
expressions, a few other things). The goal is something useful,  
usable, and very implementable (on top of current systems).

Cheers,
Bijan.

Received on Monday, 1 October 2007 08:01:47 UTC