Besides the existing support for microformats, we have already shared our plans for supporting other standards for embedding metadata into HTML. Today we are announcing the availability of eRDF metadata for SearchMonkey applications, which will soon be followed by support for RDFa. SearchMonkey applications can make direct use of the eRDF data by choosing the com.yahoo.rdf.erdf data source, while RDFa data will appear under com.yahoo.rdf.rdfa. Nothing changes in the way applications are created: as SearchMonkey applications have already been built on a triple-based model, the same applications can work on both microformat, eRDF or RDFa data.
Content publishers, however, will now have an even wider array of choice for providing metadata inside HTML. Therefore it is worthwhile to briefly summarize the key differences between microformats, eRDF and RDFa and the possible migration paths across these approaches:
- Microformats allow publishers to provide metadata about a fixed set of types. This currently includes addresses (vCard), events (hCal), reviews (hReview), feeds (hFeed) and social relations (XFN). The set of properties for each type is also fixed. We recommend the use of microformats in case the information fits one of these microformats supported by SearchMonkey. We are constantly expanding the list of supported microformats, but it is still likely that microformats will not meet all metadata needs.
- eRDF and RDFa in contrast are generic formats that can be used in combination with any RDF/OWL vocabulary and therefore can be used to represent any metadata. Unlike microformats, RDF/OWL vocabularies can also be arbitrarily mixed and extended. The use of a common or industry standard vocabulary is still recommended so that other application developers can easily interpret your metadata.
- While eRDF and RDFa look and feel much the same, there are differences in expressivity: while RDFa is complete with respect to the RDF model, not all RDF statements can be represented in eRDF. For example, in eRDF all statements have the current page as a subject (or resources defined within the current page). Further, there is no support for datatypes, which are recommended in SearchMonkey for example to provide the currency of monetary values or units.
- Migration is easy from less expressive representations towards more expressive representations, i.e. from microformats to eRDF or RDFa, or from eRDF to RDFa. Migration in the other direction is more problematic and likely result in the loss of some metadata. We suggest publishers who have currently no markup to adopt either RDFa (which will be soon picked up the Yahoo! crawler) or eRDF (which is already available from the search index). RDFa is a specification backed by the World Wide Web Consortium (W3C) and it is supported by a wide range of tools, but eRDF has been around longer - so we wanted to support both.
There are plenty of resources to familiarize with both eRDF and RDFa. The SearchMonkey guide has a brief overview of the topic. The eRDF specification and the RDFa Primer are more technical, but also complete, contain plenty of examples and still fairly easily readable. The tools supporting eRDF are listed on the same page as the specification. Here are some links to the RDFa implementations and tools.
In summary, our support for eRDF and RDFa brings even more choice for publishers while open up new data sources for application developers!
Data Architect, SearchMonkey