[uf-discuss] a very early draft proposal hTagcloud

[Drew McLellan]

On 20/9/2006, "David Janes" <davidjanes at blogmatrix.com> wrote:

>I would suggest that hTagcloud say that: "presentation is
>orthogonal to what we're trying to do -- provide semantic markup of
>tag clouds. Tag cloud implementers should come up with their own best
>way of handling the presentaion -- in fact, they probably already
>have! As a bonus, as soon as MS gets its act together you'll be able
>to do the presentation directly off the semantics, but we're not there
>yet".

I think this is an important point. Ultimately in marking up a tag cloud,
we should be representing the value of each data point.

This is what we already do with dates in hCalendar - we use an ISO
datetime to give the absolute value. Common presentation techniques are
a representation of that date in a friendly English-language format, and
plotting the date on a calendar grid.

A common presentation technique for geo is to put a pin on a map at the
corresponding location, but the underlying microformat details latitude
and longitude.

One might argue that the same should hold true of a tag cloud. It's not
inherently presentational, it's a bunch of (often relational) data
points. Those points should be equally capable of being formatted in a
data table or plotted as a pie chart. Even if the most common
presentation would be a cloud. It's all about the data.

drew.