Even with the dozens of new tags in HTML5, there is no way to provide meaning for every piece of content. This impoverishes the web: machine-readable tags that provide context make for a richer, more semantic internet, better and more accurate searches, and more powerful data aggregators. But short of an XML utopia, covering every possibility of meaning is a near-infinite task.
Instead, there are several measures that add to, or extend, the semantics of web pages:
Dublin Core meta tags, and their progenitors, attempt to summarize the context of the entire page, adding information such as authorship and revision date. However, they can only describe the page as a whole: as meta tags are written in the
<head>section, it is impossible for such tags to drill down and specify the individual content of the page.
Microformats extend tags via special values for
idattributes. Microformats have strong support by Google, Facebook and YouTube through schemas like hcard (to represent people, companies, organizations, and places), calendar (for events), recipe, and XFN (to diagram relationships between people).
To this, HTML5 adds microdata, a way of extending the meaning of tags through a shared vocabulary. Unlike meta tags, microdata is written in the body of an HTML document; unlike microformats and RDFa, it does not abuse
del, instead using new attributes that are built into HTML5 such as
item type, with established, central schemas such as schema.org and data-vocabulary.org. Google and Microsoft are both supporters of microdata, with more adopters being added.
“Which system should I use?”
Which system you use depends on what you are trying to do: