[FoRK] Blogs, mailing lists, and hammer / screw

Stephen Williams sdw at lig.net
Thu Aug 21 01:58:55 PDT 2008


Jeff Bone wrote:
>
> On Aug 20, 2008, at 11:09 PM, Stephen D. Williams wrote:
>
>> Jeff Bone wrote:
>>>
>>> On Aug 20, 2008, at 3:25 PM, Stephen D. Williams wrote:
>>>> True.  I suppose the thing that bugs me is the general lack of a 
>>>> link system, a la RDF/OWL, where I can note specific knowledge.
>>> 80/20
>> 20/80...  The way we use computers now is hideously inefficient and 
>> primitive when you step back a bit.  I have a whole monologue about 
>> it, all ready for an elevator pitch some day.
>>>
>>> IMHO, the full RDF data model (not any particular serialization) is 
>>> God's Own Data Model --- i.e., the ultimate abstract / "elemental" 
>>> data model to represent arbitrary information.  (The reification 
>>> problem in RDF, both in XML serialization and abstractly, is easily 
>>> solved with the addition of an "escaping" mechanism, which has been 
>>> solved in every embedded serialization mechanism to date --- no need 
>>> for 4-tuples everywhere, just escaping and nesting.)
>> I think that nesting is a strange solution.
>
> Nesting is usually bad, the way it's usually done --- that was the 
> root of my original complaint of folders, it implies "containment" 
> which is usually much broader than what is needed.  The kind of 
> nesting I was talking about is a kind of shorthand for logical 
> entailment in RDF, which is not exclusive.
Ahh, I thought you were talking about the RDF meta-statement method 
where you nest / quote to go meta about something:
(bob knows joe)
(mary said (bob knows joe))
(I think (mary said (bob knows Joe)))

And I have no idea what you meant by "escapes".

Containment in a hierarchy is a way of linking meta information to 
something, it is just much more restrictive than non-tree and/or 
multiple-"inheritance" methods.  Additionally, it is missing the ability 
for links between things and having a usually very restricted form of 
metadata (directory name for instance).

I think of hierarchical categorization as being equivalent to tagging 
with a similar tag path:
report -> /company1/project3/week2

The hierarchical categorization is valuable, it just needs better 
grounding in a more rich and flexible semantic space.
>
>> 4 years ago was eons ago.
>
> Indeed, but what progress has been made since?
A lot I think.  Things like this, for instance:
http://mqlx.com/~david/parallax/

sdw
>
> Seems like we're stalled...
>
> jb



More information about the FoRK mailing list