I’ve just been to XTech 2008 in Dublin, a great conference covering web development, open source, Web 2.0 and open standards.

One of the main themes that emerged was that the semantic web is already here if you know where to look - it’s appearing in parallel with the existing web, with RDFa and microformats like hCard and XFN embedded in pages. Ontologies like SIOC and FOAF are enabling semantically-rich data to be moved from one system to another, and standards like OpenId and OAuth are making it possible to provide secure access to data across api boundaries.

There are new tools that help visualise and navigate the parallel web that increasingly exists alongside older document-based data, like the Tabulator plugin for Firefox that displays RDF hidden in pages, and Google’s Social Graph api that make it easier to navigate linked data programmatically.

In terms of existing content sites, data can be immediately made more open by redesigning the site to have consistent uris and embedded RDFa (London Gazette), or in a more dynamic site like the Guardian Online, pages can be constructed automatically to pull in disparate sources of data (sports results, 3rd party content) and then referenced as persistent uris.

Data can increasingly be mixed and enriched - one interesting project used socially authored content to augment content on the BBC archive site (Yahoo for term extraction, Wikipedia for providing further information on those related topics, and then DBpedia for disambiguation - by examining each of several potential matching pages for one that contains additional terms from the original page, to confirm the context).

Other types of site are appearing like FireEagle, which are not really “sites” in the conventional sense. FireEagle is a location broker - somewhere that you can maintain and update a record of your location (determined by mobile mast locations, by wifi access point, or by setting it programmatically), which can then be used by any number of location-based services. OAuth underpins this, as a mechanism for allowing apis to be given permissions to query each other. One interesting point was that OAuth may lead to difficulties with building a strictly RESTful site - for example, a site may want uris that represent the location of a specific user, which would typically include the users’ id as part of the uri. However, using OAuth means that there may only be a token that represents the user for a given session, rather than a user name, so forming uris based on that token may not be appropriate.

One of the most exciting demonstrations of how new applications can be built from reworking data with semantic techniques was Andrew Walkingshaw’s session on Golem and CrystalEye, in which he showed how you can apply an ontology to existing data, then mine that data for relationships and re-visualise it as geographical distributions or an academic social graph.

Overall, a very impressive conference, with plenty of food for thought, and plenty of examples of how the semantic web is here around us already.