[Foundation-l] Project Proposal: Wikicat

Fri Jul 28 22:47:22 UTC 2006

On 7/29/06, Jonathan Leybovich <jleybov at yahoo.com> wrote:
> A catalog has language-specific
> data, for sure, but this is not multi-lingual data-
> the language(s) in which a book's title is
> historically expressed by the author or publisher is
> important, and you cannot just do your own translation
> into an arbitrary language and say that is also the
> book's title.

A translated version of the book should probably have its own
DefinedMeaning, since you will want to relate all kinds of information
specifically on that level. It could be linked to the original edition
not so much through the synonyms/translations, but through a special
relation type, e.g. "is translated edition of." Even with that
information, you may _still_ want to translate the title to languages
where no actual edition is available, so you can cite a book e.g. with
both its official title, and an unofficial translated title in the
language of the Wikipedia edition you're using.

> This also does not touch the performance/scalability
> issues of storing all text data, all numerica data,
> etc. in one table.

It won't be quite as simple as that, but let's discuss that.

> Regarding different referencing styles, I'm open to
> anything though I think you'll find that in practice
> standard numbers like ISBN are less cumbersome to use
> than titles.  For example, <<ref:The Davinci Code>>-
> does this mean the book, the movie, the audio book, or
> "The Davinci Code: Fact or Fiction?" ?

Referencing books that also have movie adaptations doesn't seem quite
as common. If I look at a real-world example, e.g. [[Emu]], I'll find
references like

* The heat load from solar radiation on a large, diurnally active
bird, the emu (Dromaius novaehollandiae)
* Ventilatory accommodation of oxygen demand and respiratory water
loss in a large bird, the emu (Dromaius novaehollandiae), and a
re-examination of ventilatory allometry for birds
*  Endocrine and testicular changes in a short-day seasonally breeding
bird, the emu (Dromaius novaehollandiae), in southwestern Australia.

I don't think any of these will be turned into movies soon. ;-) In
addition, you could even capture the type of reference with the
template name, e.g. <<book:The Da Vinci Code>> would only refer to an
entity which has the class membership "book". You could also do
pre-save transformations, that is, the user types <<book:Some title>>,
and if the expression unambiguously refers to one publication, a
unique identifier is automatically inserted into the reference in
addition to the title.

Of course, the ideal user interface would probably give you a little
pop-up when you click on a toolbar icon, let you search (or add!) the
reference information, and insert the right tags into the wiki source
text.

> Also, citation is not just fetching bibliographic data
> for the purposes of displaying it in an info box like
> other information.  It is fundamentally about
> associating an assertion with evidence or support, and
> so must capture the cited "text" as well as the
> paraphrase text. Here is a mock-up of these idea in
> the context of an enhanced article validation feature:
>
> http://meta.wikimedia.org/wiki/Image:Wikicite_spider_review_mockup.jpg

:-) I see we have indeed been thinking about very similar problem
areas. You'll find a mock-up of a simple scoped citation syntax in
p.190 of the first edition of my book:
http://medienrevolution.dpunkt.de/files/Medienrevolution-1.pdf

I like your systematic source review, though I'm not sure adding this
additional UI layer is necessary. One thing to keep in mind: In a wiki
review process, you'll probably almost never flag things as
"misleading" or "made-up" -- instead, you should encourage direct
editing of the content.

I think we'll have lots to talk about.

Erik