Cover page images (keys)

State of the Semantic Web

Bangalore, 23 February, 2007

Ivan Herman, W3C

What will I talk about?


Let us look at some results first!

The basics: RDF(S)

The basics: RDF(S) (cont.)

Ontologies: OWL



A mix of vocabularies/ontologies (from life sciences)…

diagram showing a large number of HC related ontologies bound via a RFD-like graph

Ontologies, Vocabularies

Querying RDF: SPARQL

SPARQL as the only interface to RDF data?

SELECT ?translator ?translationTitle ?originalTitle ?originalDate
FROM <http://…/TR_and_Translations.rdf>
   ?trans rdf:type trans:Translation;
		  trans:translationFrom ?orig;
		  trans:translator      [ contact:fullName ?translator ];
		  dc:language           "fr";
		  dc:title              ?translationTitle.
   ?orig  rdf:type rec:REC;
		  dc:date               ?originalDate;
		  dc:title              ?originalTitle.
ORDER BY ?translator ?originalDate

A word of warning on SPARQL…

Of course, not everything is so rosy…

How to get RDF data?

Data may be around already…

Data may be extracted (a.k.a. “scraped”)

Getting structured data to RDF: GRDDL

Getting structured data to RDF: RDFa

GRDDL & RDFa example: Ivan’ home page…

screen dump of Ivan's home page at W3C

…marked up with GRDDL headers…

source highlighting the GRDDL headers

…and hCard microformat tags…

source highlighting the GRDDL headers

…yielding; …

<rdf:RDF xmlns:rdf=""
   <c:Vcalendar xmlns:r=""
      <c:prodid>-// 0.6 (BETA)//EN</c:prodid>
         <c:Vevent r:about="#ac06">
            <summary xmlns="" xml:lang="en">W3C@10, 
				W3C AC Meeting and W3C Team day</summary>
            <dtstart xmlns=""
            <dtend xmlns=""
            <url xmlns=""
            <location xmlns="" xml:lang="en">Tokyo, Japan</location>
            <geo xmlns="" r:parseType="Resource">
               <r:first r:datatype="">35.670685</r:first>
               <r:rest r:parseType="Resource">
                  <r:first r:datatype="">139.770813</r:first>
                  <r:rest r:resource=""/>

(see the full file if interested…)

…marked up with RDFa tags…

source highlighting the RDFa tags

…yielding; …

<rdf:RDF xmlns:foaf="" 
		                      xmlns:rdf="" >
  <foaf:Person rdf:about="">
    <foaf:mbox rdf:resource=""/>
    <foaf:workInfoHomepage rdf:resource=""/>
    <foaf:workInfoHomepage rdf:resource=""/>
    <foaf:workInfoHomepage rdf:resource=""/>
    <foaf:name>Ivan Herman</foaf:name>
    <foaf:workplaceHomepage rdf:resource=""/>
    <foaf:schoolHomepage rdf:resource=""/>

(see the full file if interested…)

SPARQL-ing such data

          ?orgRole ?orgName ?orgHome
# Get RDFa from my home page:
# GRDDL-ing
?foafPerson  foaf:mbox ?mail;
             foaf:homepage ?home.
          ?individual  contact:mailbox ?mail;  
             contact:fullName ?name.
?orgUnit ?orgRole ?individual;  
          org:name ?orgName; 
          contact:homePage ?orgHome.

Linking to SQL

SPARQL as a unifying point?

diagram showing a sparql that can be connected to an rdf datafile, a document via grddl, and to a database via an sparql/sql bridge

Missing features, functionalities…


Some typical use cases

In an ideal World…

diagram showing star-like format of ellipses representing rule systems, all with dual arrows connected to a box stating 'full RIF format'

In the real World…

RIF “core”: only partial interchange

diagram showing star-like format of ellipses each with a yellow box core, and a separate core box in the middle to which all connect

RIF “variants”

like the core diagram, but each core is surrounded by different sized rectangles

Possible variants: F-logic, production rules, fuzzy logic systems, …; none of these have been finalized yet

Role of variants

core figure
core figures plus a cloud of rule systems core figures plus a cloud of rule systems clustered around variants with exchange arrows core figures plus a cloud of rule systems clustered around variants with exchange arrows

“Light” ontologies

Revisions of RDF and OWL?

Revision of the RDF model?

These are just ideas floating around…

Revision of OWL? (OWL 1.1)

Other items…

(Need a new PhD topic?wink smiley)

A major problem: messaging


RDF RDF/XML! (cont.)

RDF is not that complex…

Semantic Web Ontologies on the Web!

SW Ontologies some central, big ontology!


diagram showing a large number of HC related ontologies bound via a RFD-like graph

Semantic Web an academic research only!

May start with small communities

Some RDF deployment areas

Library metadata Defense Life sciences
Problem to solve? single-domain integration yes, serious data integration needs yes, connections among genetics, proteomics, clinical trials, regulatory,…
Willingness to adopt? yes: OCLC push and Dublin Core initiative yes: funded early DAML (OWL) work yes: intellectual level high, much modeling done already.
Motivation light strong very strong
Links to other library data phone calls records, etc chemistry, regulatory, medical, etc

Some RDF deployment areas (cont)

The “corporate” landscape is moving

Applications are not always very complex…

The Active Semantic Doc picture: a doctor's file with annotations

Data integration

Life Sciences (cont.)

Left side: data silos, each its own representation on a screen, with scientist interpreting; right side: same silos, converted to rdf and co, scientist doing data right away.

General approach

  1. Map the various data onto RDF
    • assign URI-s to your data
    • “mapping” may mean on-the-fly SPARQL to SQL conversion, “scraping”, etc
  2. Merge the resulting RDF graphs (with a possible help of ontologies, rules, etc, to combine the terms)
  3. Start making queries on the whole!

Example: antibodies demo

Antibodies' demo screen dump

There has been lots of R&D

MuseoSuomi Application dump Traditional Chinese medicine example dump


Vodafone screen dump

Improved Search via Ontology: GoPubMed

GoPubMed Application dump

Other Application Areas Come to the Fore


Thank you for your attention!

These slides are publicly available on:

in XHTML and PDF formats; the XHTML version has active links that you can follow