[uf-discuss] Easy book citations

[Bruce D'Arcus]

On 7/30/06, Simon Cozens <simon at simon-cozens.org> wrote:

> That sounds like an excellent way to discourage trivial accretions - people
> would only go through the process of extending the spec if they really, really
> needed it. This means you can start with something relatively simple and easy
> to get consensus on, and work from there.
>
> What's the down-side again?

The downside is that what you consider trivial I consider essential
(say archival documents), and no doubt vice versa (I really couldn't
care less about conference proceedings or patents, but you may cite
them all the time).  Moreover, there is no objective methord to fairly
measure this.

I, however, bet we all can agree that "title" is an essential property.

The practical reality is that these fora are (aside frm a few
exceptions) dominated by people from technical backgrounds, who
typically do not understand the citation needs of fields like law and
the humanities, and often do not want to try to understand.

What is thus "trivial" or falls in the "20 %" category, I would
submit, is heavily-biased. And my experience is that bias is
self-reinforcing. A lot of people still manually code citations, and
it's because the tools they have tried, and the data models they are
based on, don't fit their needs. And this is so because the people
designing them didn't consider them. Those people don't subsribe to
tech mailing lists or bother with wikis.

There are different ways to address this issue (containment, doubling
class atttributes, using generic properties like "container-title",
and "collection-title", etc.), but I really think it's critical to get
this bit right.

Bruce