Motivation

• The Semantic Web is a big container, a universal medium for data, information and knowledge exchange
• Web documents are retrieved mainly using HTTP
• Content negotiation is a difficult thing to implement (in the correct way) and to debug
• So we need a tool that helps us to debug it

Content negotiation

• Content negotiation is a HTTP mechanism that makes it possible to serve different versions of a document
• Attending to some details of the requester, such as language or format preferred
• Obiously if more than one version is available...

Content negotiation in Apache

Apache HTTP Server provides three different approaches to implement content negotiation:

1. Type Map
2. MultiViews
3. Rewrite request

Content negotiation in the Semantic Web

Common used for the Semantic Web purposes:

Content negotiation for vocabularies

• Vocabularies (and ontologies) have an important role in the Semantic Web architecture
• Due they provide logical elements to describe resources in a concrete domain
• So their publication must receive special attention
• That's the reason because the recipes document has been started

Recipes

• Best Practice Recipes for Publishing RDF Vocabularies (W3C Working Draft)
• Recipes about how to publish RDF/OWL Vocabularies using mod_rewrite
• Step-by-step instructions
• Still some open issues

Recipes (II)

• Recipe 1. Minimal configuration for a hash namespace
• Recipe 2. Minimal configuration for a slash namespace
• Recipe 3. Extended configuration for a hash namespace
• Recipe 4. Extended configuration for a slash namespace, using a single HTML document
• Recipe 5. Extended configuration for a slash namespace, using multiple HTML documents
• Recipe 6. Extended configuration for a slash namespace, using multiple HTML documents and a query service

Recipes: example recipe 3

# Rewrite rule to serve HTML content if requested
RewriteCond %{HTTP_ACCEPT} !application/rdf\+xml.*(text/html|application/xhtml\+xml)
RewriteCond %{HTTP_ACCEPT} text/html [OR]
RewriteCond %{HTTP_ACCEPT} application/xhtml\+xml [OR]
RewriteCond %{HTTP_USER_AGENT} ^Mozilla/.*
RewriteRule ^example3$ example3.html [R=303]

# Rewrite rule to serve RDF/XML content if requested
RewriteCond %{HTTP_ACCEPT} application/rdf\+xml
RewriteRule ^example3$ example3.rdf [R=303]

# Rewrite rule to serve the RDF/XML content by default 
RewriteRule ^example3$ example3.rdf [R=303]

Vapour

• Vapour, a scripting approach (written in Python) to debug content negotiation
• A web-based validator tool to check the compliance of a web site against the recipes
• It makes multiple requests to a set of URIs and runs a test suite specifically designed to check the responses of the server
• The results of these tests are showed in pretty reports

Vapour: architecture

Vapour: architecture (II)

cup

provides the web interface

teapot

is the core of the application that launches HTTP dialogs to evaluate the responses and stores the results into the RDF store

strainer

generates the reports, both in XHTML and RDF/XML

Vapour RDF model

• Vapour works on a (in-memory) RDF store
• Where it stores all tests and its results
• Just SPARQL is necessary to generated reports (both XHTML and RDF/XML)

Vapour RDF model: example (I)

An example of a test over SIOC ontology's namespace:

#req12 a earl:TestRequirement ;
  dct:hasPart a earl:Assertion ;
    earl:mode http://www.w3.org/WAI/ER/EARL/nmg-strawman#automatic ;
    earl:assertedBy #vapour ;
    earl:test #TestResponseCode200 ;
    earl:subject :rFmCDRcX137 .

:rFmCDRcX137 a earl:TestSubject ;
  dc:title "1st request while dereferencing vocabulary URI" ;
  dc:date "2008-05-27T20:36:21.318143" ;
  earl:httpRequest a http:GetRequest ;
    http:accept application/rdf+xml ;
    uri:uri http://rdfs.org/sioc/ns# .
  earl:httpResponse a http:Response ;
    http:responseCode "200" ;
    http:content-type "application/rdf+xml" .

Vapour RDF model: example (II)

Another example of a test over foaf:Person class:

:FmCDRcX51 a earl:TestSubject ;
  dc:title "1st request while dereferencing class URI" ;
  dc:date "2008-05-27T20:38:34.433181" ;
  earl:httpRequest a http:GetRequest ;
    http:accept application/rdf+xml ;
    uri:uri http://xmlns.com/foaf/0.1/Person .
  earl:httpResponse a http:Response ;
    http:responseCode "303" ;
    http:location http://xmlns.com/foaf/spec/ ;
  vapour:nextSubject :FmCDRcX52 .

:FmCDRcX52 a earl:TestSubject ;
  dc:title "2nd request while dereferencing vocabulary URI" ;
  dc:date "2008-05-27T20:36:24.318251" ;
  vapour:previousSubject :FmCDRcX51;
  earl:httpRequest a http:GetRequest ;
    http:accept application/rdf+xml ;
    uri:uri http://xmlns.com/foaf/spec/ .
  earl:httpResponse a http:Response ;
    http:responseCode "200" ;
    http:content-type "application/rdf+xml" .

Experimental results

Vapour against a list of ten popular vocabularies (retrieved on 12/Mar/2008):

Conclusions and future work

• Conclusions:
  • Content negotiation is now easier and faster to debug with Vapour
  • Scripting languages are a good way to automatize this kind of things
• Future work:
  • Cover new methods to publish RDF data, such as RDFa
  • Extend Vapour to be a Linked Data validator
  • Re-use the information in the (RDF) reports in other/s application/s