XML Reference Guide

The Semantic Web is a great idea. It defines a way in which content can be marked up so that it may be understood by various applications that can make sense of data that is currently opaque to analysis. It allows meaningful relationships to be drawn from the data by machines as well as people. But it has one tiny minor little problem.

The web already exists.

Okay, so that's not a tiny minor little problem. It's a huge, honking problem. Why? Well, for two reasons. The first reason is that millions, or perhaps billions of pages already exist, and they don't have Semantic Web information in them. The second reason is that thousands, or perhaps millions of web programmers and content authors already exists, and they don't have the Semantic Web mindset in them.

That's not to say that they don't have an appreciation for what could be achieved with the Semantic Web. Most people would love to see a world that includes Tim Berners-Lee's vision of the Web as a place where you can perform actions such as having an intelligent agent coordinate schedules and make a doctor's appointment for you, but they don't want to have to learn complicated RDF to make it happen.

Enter microformats. Microformats are, on the surface, standard ways of representing commonly published information. For example, the vCard, a standard way of exchanging "business card"-type information has been around for more than a decade, and the iCalendar standard has been around almost that long. Both of them provide a standard, interoperable way of providing information that is commonly published on the Web. If an easy -- and most of all, standard -- way can be found for developers and content authors to represent this information in web pages, it can be available for semantic searches.

Note the use of the lowercase "s" in "semantic". Tantek Çelik has been giving talks on microformats and calling it the "semantic web", emphasizing the lowercase nature of it. This is not the Semantic Web of RDF and OWL, where the emphasis is on making data machine-readable. Instead, the emphasis is all in making data that is human readable first, and machine-readable second.

In essence, the microformats movement has two aspects. The first is to define the information represented by a particular kind of entity. For example, a vCard has the following information:

BEGIN:VCARD
VERSION:3.0
N:Chase;Nicholas
FN:Nicholas Chase
URL:http://nicholaschase.com/
ORG:InformIT
END:VCARD

The hCard microformat takes this information and provides a standard way to represent it in XHTML:

<div class="vcard">
 <a class="url fn" href="http://nicholaschase.com/">Nicholas Chase</a>

 <div class="org">InformIT</div>
</div>

Note the use of classes to specify information. This is a feature of microformats; they try very hard not to change what people are doing, but rather to simply add on in a way that is useful. For example, because this is XHTML with class information, you can simply create a style sheet such as:

<style type="text/css">
   .vcard {border: 1px dotted black}
   .fn {font-weight: bold; text-decoration: none}
   .org {display: inline}
</style>

And the hCard would be displayed as:

Adding information this way is not onerous to content authors, but it provides enough information that an aggregator could easily find specific hCard data.

And that's the other focus of microformats: defining how this information should be represented. The Microformats wiki defines them as:

simple conventions 
for embedding semantic markup 
            for a specific problem domain 
in human-readable (X)HTML/XML documents, Atom/RSS feeds, and "plain" XML 
            that normalize existing content usage patterns 
            using brief, descriptive class names 
            often based on existing interoperable standards 
to enable decentralized development 
            of resources, tools, and services 

And it does seem to be catching on. The wiki lists nine specifications, and at least as many drafts. They fall into two categories: elemental microformats and compound microformats.

Elemental microformats are meant to be composed into compound microformats. For example, the list of elemental microformats includes:

XOXO:

This microformat represents lists. In fact, to the naked eye it is indistinguishable from XHTML lists and definitions, encompassing the ol, ul, li, dl, dt, and dd elements. All are used in their normal manner, but since most people don't use the dl, dt, and dd elements properly, it is a formal declaration that they are to be used as a definition lists, definition terms, and definition descriptions, respectively. (For example, this list is marked up as a definition list.) Also, by adding the xoxo class to lists intended to be part of a microformat, you specify their purpose, and not just their presentation.

RelTag:

This microformat has a leg up in the blogging world, because of its adoption by Technorati, where Tantek Celik is Chief Technologist. It specifies that information in is a "keyword" or, more commonly, a "tag". For example, I might tag this document as:

<a href="http://www.technorati.com/tag/microformats" rel="tag">microformats</a>

to indicate that this was a document about microformats. The rel tag is what specifies that as a tag, and the href attribute is the "tag space", with the tag consisting of the last part of the URL. So while most people use URLs that point back to, say, Technorati, which maintains a listing of pages for each tag, it is only important to point to something relevant. For example, as the specification suggests, we could point to

<a href="http://www.wikipedia.com/microformats" rel="tag">microformats</a>

and still have are relevant tag.

XFN:

This is an interesting little specification that enables you to note your relationship to other people in social networks. I could create a blogroll and specify my relationship with the authors with regards to several axes, such as friendship, whether or not I've met them, geographical, professional, family, and so on. For example:

<a href="http://www.chaosmagnet.com" rel="self">Chaos Magnet</a><br />

<a href=" http://www.squidoo.com/genealogyhowtoinfo" rel="spouse">Sarah Chase</a><br />
<a href="http://www.backstopmedia.com" rel="co-worker friend met">Troy Mott</a><br />
<a href="http://www.workbench.com" rel="acquaintance colleague">Rogers Cadenhead</a><br />

RelLicense:

This format enables you to link to the license for a particular page. For example, my blog is subject to a Creative Commons license, so I could add a link of:

This weblog is licensed under a  <a href="http://creativecommons.org/licenses/by-nd/1.0/" 
rel="license">Creative Commons License</a>

The Microformats group has also specified several compound formats:

hCard:

This is the format we described at the start of this entry.

hCalendar:

Used to mark up events, appointments, and other such calendar-related information, this format encodes the information normally found in the iCalendar specification as XHTML. For example, if I wanted to add a conference on speaking and in June to my blog, I could mark it up as:

<div class="vcalendar">
<div class="vevent">
 <a class="url" href="http://www-306.ibm.com/software/rational/events/rsdc2006/">

  <span class="summary">IBM Rational Software Development Conference 2006</span>: 
  <abbr class="dtstart" title="2006-06-04">June 4</abbr>-
  <abbr class="dtend" title="2006-06-09">8</abbr>,
 at the <span class="location">Walt Disney World Swan and Dolphin Resort Orlando, Florida</span>

 </a>
</div>
<div class="vevent">
...
</div>
</div>

Note that the vCalendar div is optional; you'll want to include it, however, if you're showing more than one event. Also note that it is only the class names that are important. You can just as easily include the event in a span, and list several of them on one line. Note also that the dtend value refers to the day after the event, hence the disconnect between the data and the human readable version.

VoteLinks:

The specification provides a way to link to something without showing your approval of it. Consider this example taken from the specification:

<a rev="vote-for" href="http://ragingcow.blogspot.com"  
   title="neat spoof">Raging Cow</a>
<a rev="vote-for" href="http://ragingcow.blogspot.com"  
   title="the boycott site">I support the Raging Cow</a>

<a rev="vote-against" href="http://ragingcow.com"  
   title="nasty corn syrup drink">Raging Cow</a>

The page links to both favored an unfavored sites, using the rev attribute to distinguish between them. Yes, this is a real attribute for this element, intended to denote a "reverse link", but rarely used.

hReview:

Reviews are perhaps one of the most commonly cited purposes for microformats. This is probably because so much blogging is done to tell people what you think about something. This specification is intentionally minimalist, capturing only properties that are common to all reviews. Consider this example from the specification:

<div class="hreview">
 <span><span class="rating">5</span> out of 5 stars</span>
 <h4 class="summary">Crepes on Cole is awesome</h4>

 <span class="reviewer vcard">Reviewer: <span class="fn">Tantek</span> - 
 <abbr class="dtreviewed" title="20050418T2300-0700">April 18, 2005</abbr></span>
 <div class="description item vcard"><p>

  <span class="fn org">Crepes on Cole</span> is one of the best little 
  creperies in <span class="adr"><span class="locality">San Francisco</span></span>.
  Excellent food and service. Plenty of tables in a variety of sizes 
  for parties large and small.  Window seating makes for excellent 
  people watching to/from the N-Judah which stops right outside.  
  I've had many fun social gatherings here, as well as gotten 
  plenty of work done thanks to neighborhood WiFi.
 </p></div>

 <p>Visit date: <span>April 2005</span></p>
 <p>Food eaten: <span>Florentine crepe</span></p>
</div>

Notice that the hReview specification makes use of other microformats, such as hCard. This is, after all, a compound specification, and does include other attributes for specific types of reviews, such as product reviews or "multidimensional" reviews. See the specification for more information.

Now we need an easy way for authors to generate this information.