W3C killing off XHTML2 in favor of HTML5: bad for the Semantic Web?

As a practical matter, HTML5 looks good for writing human facing next generation web applications with multimedia support and more intuitive elements like <header>, <nav>, <section>, <footer>, etc.

The problem that I have with the W3C's decision (assuming that I understand it correctly) is that at least in my opinion the value of the web goes way beyond supporting manual web browsing and enjoying digital media assets. I think that the web should evolve into a ubiquitous decision support system - this needs software agents that can help you no matter who's computer you may be using, what type of small device (phone, web pad) you may be using, etc. In this context, decision support means help in making dozens of decisions each day. User specific information filters, search agents, and personalized information repositories will require machine readable data with well defined semantics.

One approach is to have content management systems like Drupal and Plone publish information in parallel, both:
  • HTML5 web pages for human consumption
  • RDF/RDFS/RDFS+/OWL for consumption by software agents
It is very easy (like a few lines of Ruby) to convert either entire databases or subsets from SQL queries to RDF and since many web pages are created from information in relational databases, it might be OK to use this "dual publishing" scheme. Similarly, data in RDF repositories can be used, instead of relational databases, to publish web pages. However, I would prefer generating one format of web page with semantic information embedded as, for example, RDFa and micro formats.

My vision of the web is an increasing amount of data and information that we need customizable software agents or interfaces to cull out just what an individual user needs.

HTML5 needs a well designed notation for embedding extensible semantic information that does not rely on XML's extensibility.


  1. Hi Mark, You can embed RDF in the XML serialization of HTML5.

    There is also a Microdata section that you might find interesting.

  2. Hello Chris,

    Embedding RDF in the XML serialization looks better than the microdata support. Is it a good assumption that all (most?) browsers will process the XML serialization? Assuming that the answer to this question is yes, then I guess that there is no real problem.

    I think that the reversed DNS identifiers used in microdata markup look really odd, but I could get used to it. It is not clear to me why people would not simply use a URL as an itemprop attribute referring to, for example, an RDFS resource defining types.



Post a Comment

Popular posts from this blog

Custom built SBCL and using spaCy and TensorFlow in Common Lisp

I have tried to take advantage of extra time during the COVID-19 pandemic

GANs and other deep learning models for cooking recipes