Re: [xml-dev] Six Reasons Not to use XML Attributes

["Christopher R. Maden" <crism@maden.org>]

On 03/02/2012 01:44 PM, Andrew Welch wrote:
>> That's what I meant by the field needing to be repeatable.  Chris 
>> Maden devised the system that $EMPLOYER is using, which involves 
>> personNameProfile elements which are themselves repeatable, and
>> which have children for purpose, pattern (what order to put the
>> textual elements in), the (meta-)name of this profile, full name,
>> and five elements for name parts: givenName (repeatable),
>> familyName (repeatable), prefix, suffix, nickname, and full name,
>> all of which are optional.  There is continual pressure to add
>> middleName, but so far we have effectively resisted it: canonical
>> U.S. middle names are additional given names.
> 
> Straying off topic - why not use the foaf markup?

The use case at $EXEMPLOYER[*] involved an entity[†] mastering system;
the XML serialization was primary in human-readable documents but also
interchange to and from the entity master.  That and some other business
dictates put constraints on the nature of the markup.

We ended up with a full name, which is an unstructured string, for when
the structure is not (yet) known, as well as a set of strings with
metadata (as John describes) for when the structure is known.  It’s
important to know the family name, for instance, when doing entity
recognition against a news article or legislative or court transcript;
it’s also important to know nicknames and common misspellings for
similar reasons.

The resulting markup is mappable to and from FOAF, but FOAF wasn’t
directly usable as-is.

~Chris

[*] A brief visit to LinkedIn should render this obfuscation trivial, if
anyone cares.
[†] “Entity” in the sense of real-world concept, not in the XML sense(s).
-- 
Chris Maden, text nerd  <URL: http://crism.maden.org/ >
“Be wary of great leaders.  Hope that there are many, many small
 leaders.” — Pete Seeger