RDFa and SEO

16 Dec 2009

I did a couple of talks at SemTech 2009 this year, and although I managed to upload my talk on RDFa: The Semantic Web's Missing Link, I never got round to uploading the slides for a panel discussion I was involved in, called A Shift in SEO.

I was reminded of this omission earlier this week, when a discussion around RDFa and SEO started to take off, fuelled largely by the energetic Martin Hepp (see GoodRelations & RDFa for Deep Comparison Shopping on a Web Scale and Developments in Information Retrieval on the Web from SES Chicago '09).

Since I made some points in the panel discussion that complement some of those that are being made now, I thought 'better late than never', and uploaded the slides. (Especially since it was quite smart of the conference organisers to place an SEO session into the middle of a semantic technology conference.)

My comments in the session looked at:

  1. how RDFa can help to improve the display of search results;

  2. how it can also be used to improve the accuracy of those results;

  3. and consequently, how RDFa could help to create a series of vertical search engines.

Improving display of results

This is the area that has probably seen most discussion recently, in the context of Yahoo!'s enhanced results, and Google's rich snippets. Both techniques improve the display of search results by bringing to the fore information that is specific to the item being displayed. For example, if the search result is a film, then the kind of information that should be emphasised would be whether reviewers liked the film or not, the running time, its rating (i.e., whether it's suitable for children, or not), which cinemas are showing the film near you, and so on.

The benefits to the search engine of doing this, are that users can get more done, on their site, making them more likely to return. The benefits for the companies providing the data are improved click-through.

This last point is worth stressing, since many people on the outside of SEO assume that it's all about trying to get a particular web-site to appear as high in the listings as possible, but for many sites, click-through is probably more important than ranking. Some SEO experts at the conference were saying that even if adding RDFa to a site gave an improvement in click-through of only a couple of percent, sites would see that as worth the effort -- yet as Peter Mika said in Year of the Monkey: Lessons from the first year of SearchMonkey, adding RDFa and Microformats to a page gives significantly better click-through than a mere few percent.

Hence the interest in Martin's GoodRelations ontology.

Improved search accuracy

Search accuracy is important in this, too. Obviously it's beneficial for the users of search engines, in that it can help them to find the information they want, faster. But it's also significant for site creators and SEO practitioners, because it means that sites are increasingly found in the right place.

Once again, this seems to be at odds with the general view of SEO, but an increasing part of the SEO job description is the writing of relevant articles on topics that relate to your products and services, as a way to bring people to your site. Since the search engines are increasingly clever enough to differentiate between a bunch of keywords dumped into a page, and an article with real content, a virtuous circle is created, rewarding 'proper' articles with improved rankings.

RDFa can help this further, because it allows authors to make their pages unambiguous. And if the search engines are rewarding accuracy of ordinary text pages, it makes sense that they'll reward accuracy even more, if those pages contain RDFa.

Vertical search engines

The final point I looked at was the way that RDFa will allow the search giants to offer partitioned search engines, aimed at particular audiences.

It's true that there are many search engines already available, for specialist areas, but most of them tend to be out of date, or missing information altogether, and they often require bloggers to register their site for crawling.

The major search engines are often crawling these sites already anyway, but page ranking algorithms will hide them away in the 1000th page of your search results. By adding targeted mark-up to web-pages it becomes easier for search engines to differentiate the subject-matter of the pages, and so offer specialised views on their data.

Conclusion

RDFa is already impacting search in positive ways, and SEO is adapting accordingly. But there is a great deal further that search and SEO can go towards being more 'semantic', and RDFa has a lot to offer in realising the potential.