Geospatial Semantic Web Blog

GRDDL (Gleaning Resource Description from Dialects of Languages) is now a W3C Recommendation. Some people prefer RDFa over GRDDL, but I think they are simply two different languages for solving similar problems with distinctive domain constraints.

GRDDL Basics

GRDDL attempts to solve an ago-old Semantic Web problem: how to publish semantic markups on the Web. The basic idea behind GRDDL is as follows. XHTML and XML are popular languages for publishing information on the Web. At present it’s difficult for computer programs to process semantic information that is hidden in XHTML and XML documents. RDF is a language suitable for describing semantic information. GRDDL defines how content information described XHTML and XML can be bootstrapped into RDF.

Central to GRDDL is the use of XSL Transformations. Publishers who follow GRDDL conventions will carefully design their web documents in a such way that it will be possible for computer programs to download XML stylesheets and use them to extract RDF from the the original XML and XHTML documents. For an illustration, see GRDDL Primer.

GRDDL for KML and GeoRSS

Many people (including myself) think GRDDL has great potential in turning existing XML and XHTML documents into a web of semantic documents. Several use cases have been described in the GRDDL Use Cases document, which include use cases for health care, meeting scheduling, wiki and e-learning, social networks, and digital content management.

Some people are skeptical about GRDDL because it’s an approach that requires significant amount of design and technology implementation. It also suffers problems that are inherent in XSLT technology. For example, documents designed according to the GRDDL standard may be “brittle” because small errors in either the associated stylesheets or the documents’ content can break the whole RDF bootstrapping process.

I agree GRDDL is a more brittle approach comparing to RDFa, which directly embeds RDF in web documents, but I think GRDDL has its own advantages. First, minimal changes are required to make existing XML and XHTML documents readable by RDF robots. Second, it imposes less overhead in bootstrapping documents that use standard vocabularies (e.g., RSS, GeoRSS, KML). Third, it separates the content (i.e., what the document is about) from the bootstrapping logics (i.e., how to transform the document).

Here is an example. Let’s say Google decided to embrace RDF in KML and GeoRSS. Using GRDDL, Google Maps team defined standard stylesheets for bootstrapping RDF statements from KML and GeoRSS documents. Once they were defined, to get feeds readable by RDF-robots, web publishers just had to add few lines of GRDDL markups in their feeds.

Concluding Remarks

I’m happy to see GRDDL is now a W3C Recommendation. Hopefully, more companies and organizations will embrace this new standard. It will help clients to bootstrap RDF from the existing XML and XHTML documents. I don’t think GRDDL is in any competitions with alternative standards such RDFa. They are simply two different languages for solving similar problems with distinctive domain constraints.