O'Really?

The Web Ontology Language (OWL) is a language for creating ontologies on the Web. It does exactly what it says on the tin. But what is an ontology? One way to think of it is as a better way of storing data and knowledge. Instead of just capturing and describing data in a databases, ontology languages like OWL provide ways to capture and describe knowledge in a knowledge base. Ontologies can allow more intelligent querying, integration and understanding of data than is possible using a plain old relational database.

Since 2003 developers and users of the Web Ontology Language, abbreviated to OWL (not WOL), have been gathering at a two-day workshop called OWLED (OWL Experiences and Directions). This year the workshop is in Karlsruhe in Germany. The full list of accepted papers is available, as with previous years, this years workshop has a distinctly biological flavour to the proceedings:

• Biological Situational Modeling: Defining Molecular Roles in Pathways and Reactions. Michel Dumontier.
• GO faster ChEBI with Reasonable Biochemistry, yours truly, see also the poster version
• Use of OWL 2 to Represent a Biomedical Knowledge Base Extracted from the GENIA Corpus. Rafal Rak, Lukasz Kurgan and Marek Reformat.
• The OWL of Biomedical Investigations. Melanie Courtot, William Bug, Frank Gibson, Allyson L. Lister, James Malone, Daniel Schober, Ryan Brinkman and Alan Ruttenberg.
• A Call for an Abducting-Reasoning Feature in OWL-Reasoning Tools toward Ontology Quality Control. Michael Bada, Chris Mungall and Lawrence Hunter.

After the paper presentations there was a panel session:

How Might OWL Fail?

Similar to tomorrows panel at the International Semantic Web Conference (ISWC 2008 ) An OWL 2 Far?*, OWLED 2008 closed with a session discussing how OWL might fail, based around three questions.

1. What single thing should the OWL community stop doing as it is a barrier to adoption?
2. What single thing should the OWL community start doing in order to foster adoption?
3. Playing the role of an evil mastermind, if you really wanted to derail OWL’s progress and success, what would you do?

This was run by Michel Dumontier, Tom Heath (Talis), Nick Drummond, Carsten Lutz and chaired by Sean Bechhofer. Some responses below, in note form (life is too short).

Michel Dumontier: OWL is a tangent to RDF, the  logic has to “disappear” so people can just do their jobs. We can foster adoption by building a “masterplan”, envisioning specific applications five years down the road, more evangelical and outreach work. OWL community needs to diversify. Wow moment, was not having to write SQL queries – this is a major selling point for OWL – often preferable to dealing with SPARQL.

Carsten Lutz: what would it mean for owl to fail? reasoning is a distinguishing feature. industry will want a mature scalable technology. too much feature bloat? adding more expressive power to the languages at the expense of fast and efficient reasoning. LOOM failed for this reason.

Tom Heath: experienced semantic web developer, doesn’t understand OWL. Fear, Uncertainty and Doubt (FUD) about OWL. don’t need to do anything, How might owl succeed? Marketing issue, appeal to semantic web community and wider web community of mashup developers. What is the “elevator pitch” for OWL? This is far from clear.

Nick Drummond: OWL community is unapproachable,  Google hits for “owl+ontology” can be inaccessible, need much more accessible points for developers. A demonstrator application – “killer app(s)”. To scupper OWL would cut all funding for tools to sabotage them. OWL is relatively successful in niche markets (e.g. biomedical ontologies), but less successful on the wider web and can’t deal with lots of instances.

Alan Ruttenberg: Big selling point for OWL is “nonsense detection” and data integration (similar to the linked data effort). Related semantic web efforts have “lots of data, not so many questions”. Common logic, as a rival also SKOS (“dumbing down”?), JavaScript reasoner required to widen user base (e.g. like Tabulator), need a proper querying language.

Ian Horrocks: difficult for OWL to compete with languages that allow you to say anything inconsistent, with uncertain implications, all very quickly (!). See RDF to OWL: The Making of a Web Ontology Language.

Alan Rector: OWL competes with UML to some degree (other people disagreed with this), which is very well tooled. OWL community needs to figure out how it interfaces with UML.

Conclusion

There is a small but active community of people using and developing the Web Ontology Language (OWL) to represent and reason about different kinds of biomedical knowledge. Personally, it is my first time at OWLED, so it has been good to meet some new people: Rafal Rak (from Canada), Tania Tudorache and Timothy Redmond (developers of Protégé from Stanford, California), Biochemist Michel Dumontier from Carleton University and Eric Prud’hommeaux from the World Wide Web Consortium (W3C). If you are interested in participating next time around, OWLED will be in Washington D.C. in 2009.

* “An OWL 2 Far?” is a dreadful (but amusing) pun on “2 / too” in OWL version 2.0, [1] perhaps we should beware of version 2.0 syndrome?. See also OWLED’08 in brief by Marijke Keet and What is OWL good for? by Sandro Hawke.

References

1. Bernardo Cuenca Grau, Ian Horrocks, Boris Motik, Bijan Parsia, Peter Patel-Schneider, Urike Sattler (2008). OWL 2: The next step for OWL Journal of Web Semantics (in press) DOI:10.1016/j.websem.2008.05.001
2. Ian Horrocks, Peter Patel-Schneider and Frank van Harmelen (2003). From shiq and rdf to owl: The making of a web ontology language. Journal of Web Semantics 1 (1), 7-26. DOI:10.1016/j.websem.2003.07.001
3. Video lectures An OWL 2 far?

[Creative Commons licensed picture of Great Grey OWL (Strix nebulosa) by Brian Scott]