Heuristic Tests for Data Tables (Discussion)

Sadly, there are tables where <td> is used for everything. For examples:

* <http://sports.espn.go.com/nhl/players/profile?statsId=1339>
* <http://www.fleetlions.org.uk/events.html>
* <http://esw.w3.org/topic/RecentChanges>

More examples easily found from web searches like 
<http://www.google.com/search?q=table+2>:

* 
<http://www.osha.gov/pls/oshaweb/owadisp.show_document?p_id=9993&p_table=STANDARDS>
* <http://www.biomedcentral.com/1471-2105/6/60/table/T2>

HTML4 does not try and guess header relationships. Doing this in HTML5 would 
be a new ability for HTML. <td>-only data tables are about as common as 
tables which <th>, in those I've seen. This ability, if possible, would add 
significant accessibility to these present-day tables with no authoring 
effort.

In the first example from ESPN:

1. The first row is several columns wide and spans the entire table. It 
contains a text string. You could imply this is a <caption>.
2. The data types of the second row are all text strings with no numbers. 
This is different to the subsequent rows. You could imply these are <th> 
cells for those subsequent rows.
3. In those subsequent rows, several formats are used. These formats are 
fairly consistent down each column. You could imply these are the <td> cells 
which correspond to the <th> cells you implied in step 2, without 
associating them to the implied <caption> from step 1.

I don't have a concrete proposal for what these tests should be. The above 
steps might not work in other tables.

You'd need to determine which tables are data tables and which are layout 
tables, too. From what I've read, screen readers are already quite capable 
at that. It would be cool to get them more involved in this.

Ideas and research are welcome. :-)

--
Ben 'Cerbera' Millard
Collections of Interesting Data Tables
<http://sitesurgeon.co.uk/tables/readme.html> 

Received on Tuesday, 14 August 2007 17:34:21 UTC