Re: Contexts (spinoff from copy and wrap rdf statements)

Danny Ayers wrote:

>>>At 1:04 PM -0800 2002-11-21, Seth Russell wrote:
>>>
>>>      
>>>
>>>>I totally agree.  I really don't understand why we can't just agree to
>>>>refer to the set of triples (the model) encoded by the document at
>>>><urlA> as <urlA#ThisGraph>.
>>>>        
>>>>
>>>What about graphs that are not encoded by documents with URIs?
>>>      
>>>
>>Ok, take for example the graph of all RSS items that contain the key
>>word phrases "Star Trek", "uiversal translator", and "GroupFormingHere".
>> Granted there is no particular document that encodes that graph at
>>2:15 AM on 11/22/2002.  But it is still our intention to find this graph
>>and give it a URI so we can talk about it, right?   Is that your
>>question?
>>    
>>
>
>Hmm - if I want to use my foaf data elsewhere I can presently refer to it
>using the URL of a file that contains it. This is the de facto URI for the
>graph containing that information. But following the suggestion above,
>wouldn't I have to use another URI:
>
>http://somewhere/me.rdf#http://somewhere/me.rdf
>
>Seems a little strange somehow...
>
No, that wasn't what i wanted to imply.    I think TimBl would say that 
 <http://somewhere/me.rdf>  identifies a document.  But a document is 
not an abstract graph.  So if that URI identifies a document, what URI 
identifies the actual graph ?   I'm just proposing a standard way of 
distinguishing between a document and the set of triples which it 
encodes.   If I copied your foaf document to this email it would have a 
different document URI, but the triples in the graph would be the same. 
 So the triples in the email document would be isomorphic with the 
triples in the web document.  We should be able to say that  ...

<http://somewhere/me.rdf#ThisGraph>
    sameGraphAs <mid:foo#ThisGraph>.

So the graph itself has an identity and a URIREF.  

David raised an important question that there are some graphs which have 
no document to which we can affix the #ThisGraph and therefore  those 
graphs (even though they might exist) would have no URI.   I think that 
aggregations of triples from different sources by some group of people 
fall into this catagory.   But my claim would be that a group of people 
can always assign some kind of URI to that aggregation.  Once we have 
that URI, we also have the URIREF of the set of triples.

>(btw, you can add another item to your graph:
>http://ideagraph.net/rss/Blogfeed.pl#http://ideagraph.net/rss/Blogfeed.pl
>)
>
Well I shouldn't have to add it myself ... the fact that you asserted 
that item in your blog should have entailed it's assertion in that graph.  

In other words:
[1]   Your  item #84894783 contained the  word "GroupFormingHere"
[2]   There is a graph of all items containing that word
--entails--->
<http://www.citnames.com/blog/archive/2002_11_17_index.html#84894783>
    isAssertedTo 
<http://purl.org/context?keywords=GroupFormingHere#ThisGraph>.

Seth Russell
http://radio.weblogs.com/0113759/