Re: restlets & rdf

[Story Henry]

Ok, I found that Tim Berner's Lee and co, are using their own ontology which is similar to the http in rdf one I mentioned earlier, with the advantage that it is being deployed in a real experimental application that does exactly what I need to do too.

Try the latest development version of the tabulator with Firefox at
http://www.w3.org/2005/ajar/tab


http://bblfish.net/people/henry/card
in the box at the top, then pressing Go!




and clicking the Data as N3 view. I get this:

@prefix http: <http://www.w3.org/2007/ont/http#>.

[ http:location
   "http://bblfish.net/people/henry/card.n3";;
http:status
   "303";
http:statusText
   "See Other" ].

The tabulator is written in javascript and contains a quad store (it knows where it got the triples from). You can see how it keeps track of the http requests by typing in a url such as Tabulator will then fetch the information, which might involve redirects etc. You can then view how it stores this metadata by clicking the little wheel, clicking on the "access" relation The other way of seeing this (some of this information) is to go to the bottom of the tabulator page, open the Log section, and click "dump store to log" right at the bottom. This dumps the store in Ntriples format. (so there must be some info missing, as it does not keep the context). But this is also very helpful. <http://bblfish.net/people/henry/card> <http://www.w3.org/2007/ont/link#request > _:n70 . _:n70 <http://www.w3.org/2000/01/rdf-schema#label> "Access of http://bblfish.net/people/henry/card " .

_:n70 <http://www.w3.org/2007/ont/link#handler> _:n71 .
_:n70 <http://www.w3.org/2007/ont/link#status> _:n69 .
_:n70 <http://www.w3.org/2007/ont/http#status> "303" .
_:n70 <http://www.w3.org/2007/ont/http#statusText> "See Other" .

_:n70 <http://www.w3.org/2007/ont/http#location> "http://bblfish.net/people/henry/card.n3 " .

#( there is no more info on _:n71 or _:n69 presumably because of the 303. ) <http://bblfish.net/people/henry/card.n3> <http://www.w3.org/2007/ont/link#requestedBy > <http://bblfish.net/people/henry/card> .

#( notice above that it keeps track of how requests are followed through ) _:n0 <http://www.w3.org/2007/ont/link#source> <http://bblfish.net/people/henry/card.n3 > . <http://bblfish.net/people/henry/card.n3> <http://www.w3.org/2007/ont/link#request > _:n73 . _:n73 <http://www.w3.org/2000/01/rdf-schema#label> "Access of http://bblfish.net/people/henry/card.n3 " .

_:n73 <http://www.w3.org/2007/ont/link#handler> _:n74 .
_:n73 <http://www.w3.org/2007/ont/link#status> _:n72 .
_:n70 <http://www.w3.org/2006/link#redirectedRequest> _:n73 .
_:n73 <http://www.w3.org/2007/ont/http#status> "200" .
_:n73 <http://www.w3.org/2007/ont/http#statusText> "OK" .

_:n73 <http://www.w3.org/2007/ont/httph#date> "Wed, 21 Nov 2007 16:14:08 GMT" . _:n73 <http://www.w3.org/2007/ont/httph#server> "Apache/2.0.55 (Unix) DAV/2 mod_perl/2.0.2 Perl/v5.8.4" . _:n73 <http://www.w3.org/2007/ont/httph#last-modified> "Tue, 04 Sep 2007 14:53:46 GMT" .

_:n73 <http://www.w3.org/2007/ont/httph#etag> "\"4cf-15f5-74c3d280\"" .
_:n73 <http://www.w3.org/2007/ont/httph#accept-ranges> "bytes" .
_:n73 <http://www.w3.org/2007/ont/httph#content-length> "5621" .

_:n73 <http://www.w3.org/2007/ont/httph#content-type> "text/rdf+n3; charset=utf-8" .

#( It also adds some information as to the type of a resource )

<http://bblfish.net/people/henry/card.n3> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type > <http://www.w3.org/2007/ont/link#Document> . <http://bblfish.net/people/henry/card.n3> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type > <http://www.w3.org/2007/ont/link#TextDocument> .

Henry

On 21 Nov 2007, at 11:18, Story Henry wrote:

I think this can make a helpful case as to why one should keep track of this information. It should be no trouble extracting this from RESTlets I imagine. But I thought the RESTlet group might find this insight quite interesting.

Hi,


some interesting use cases in mind:

  - Reporting test results
  - Conformance claims
  - Developing Web Apps

http://www.w3.org/TR/HTTP-in-RDF/


at the model being used here.


be more efficient for DB storage.



If I run the following curl

[EMAIL PROTECTED]:0$ curl -I http://xmlns.com/foaf/0.1/knows --trace-ascii

I get the following trace

<curl-output.txt>


which would result in the following graph in N3

@prefix http: <http://www.w3.org/2006/http#> .
@prefix http-header: <http://www.w3.org/2006/http-header#> .

_:r1 a http:HeadRequest;
  http:abs_path "/foaf/0.1/knows";
  http:version "1.1";
  http:header [ http:fieldName http-header:User-Agent;

  http:header [ http:fieldName http-header:Host;
                http:fieldValue "xmlns.com" ];
  http:header [ http:fieldName http-header:Accept;
                http:fieldValue "*/*" ];
  http:response [ http:version "1.1";
                  http:responseCode http:303;
                  http:header [ http:fieldName http-header:Date;

                  http:header [ http:fieldName http-header:Server;

                  http:header [ http:fieldName http-header:Location;

I have found an interesting ontology for HTTP called "Http in RDF". It is being developed with This could be used in the case of RESTlets to log all information about connections into an RDF database, which could then be queried using SPARQL, or used in other ways... It can also be interesting just to look The ontology is pure RDF which means that relations are very flexible. It is possible for example for relations to point to literals just as easily as it is for them to objects point to objects. Clearly for particular applications it would help to use the ontology in a more OWL-DL like manner, which would fit java use better, and perhaps even What do the resulting relations look like? Here is an example I put together to help me grok it. http:fieldValue "curl/7.16.3 (powerpc-apple- darwin9.0) libcurl/7.16.3 OpenSSL/0.9.7l zlib/1.2.3" ]; http:fieldValue "Wed, 21 Nov 2007 00:14:38 GMT" ]; http:fieldValue "Apache/1.3.37 (Unix) mod_throttle/3.1.2 DAV/1.0.3 mod_fastcgi/2.4.2 mod_gzip/ 1.3.26.1a PHP/4.4.7 mod_ssl/2.8.22 OpenSSL/0.9.7e" ]; http:fieldValue "http://xmlns.com/foaf/spec/ " ]; http:header [ http:fieldName http-header:Content- Type; http:fieldValue [ http:elementName "text/html"; http:param [ http:paramName "charset"; http:paramValue "iso-8859-1" ]

                                                 ]
                              ]
                ] .



request to the connection it is part of

_:c1 a http:Connection;
  http:connectionAuthority "xmlns.com:80";
  http:request _:r1 .

If we wanted to follow up with this with a new request to 
http://xmlns.com/foaf/spec/
then we would send another request, perhaps using the same connection

_:c1 http:request [ a http:GetRequest;
                   http:abs_path "/foaf/0.1/knows";
                   http:version "1.1";
                   # etc...

                 ] .




the resource <http://xmlns.com/foaf/0.1/knows> to its representations

Notice that the above does not tell us what the request was issued against. So one has to link the http:response [ http:body <file:///cache/sdfsdfs.rdf >; So it is clear from the above that we have a vocabulary here that is very close to the mechanics of HTTP. It may be helpful to have a few relations a little higher in the stack of abstractions to link Anyway, any thoughts on how close this model is to the Restlets model? Does it help? Using @rdf annotations from so(m)mer (https://sommer.dev.java.net) one may even be able to link the two models

quite easily .


   Henry


condensed down to 1 pointer.


Home page: http://bblfish.net/
Sun Blog: http://blogs.sun.com/bblfish/
Foaf name: http://bblfish.net/people/henry/card#me

Ps. The above may look very verbose, but remember that URLs are pointers, and so every use of a URL can be

smime.p7s
Description: S/MIME cryptographic signature