On 7/6/07, Benjamin Esham bdesham@gmail.com wrote:
I suppose my proposal—though as I said, this is really just brainstorming—is to replace the giant <textarea> with some semblance of structured data entry. I'll admit, I'm not particularly enamored with the idea of a thousand input fields on each page, but compelling users to put the license information in *this* box, and the author in *that* box, and the categories *over there* is the only surefire way to standardize the look of media pages on Commons. (A significant bonus here is that having the data structured makes it infinitely easier to search—such structure could put the Commons with the likes of Flickr as far as the Semantic Web goes.) As many people have noted on this thread and elsewhere, MediaWiki is primarily an encyclopedia engine, and needs significant changes to be really useful for media.
On one of the CCC meetings in Berlin (2005?) I hacked a little thing that would, on clicking "edit", split the page text (here:image description) in two: 1. Text (here:{{Information}}) 2. Metadata (categories, language links, non-position-dependent templates) and show them as two separate textareas (a large and a small one). Upon saving, it would append the second to the first one. That allowed for the separate editing of metadata while not interfering with any other part of the software.
It was a quick'n'dirty hack, and it was probably removed from MediaWiki in the meantime. However, something along these lines (showing only one or the other textarea) might be feasible, if properly planned and watched by many eyeballs.
As long as I'm dreaming out loud... a couple of years back there was a poll to determine whether images should be added to articles, categories, or both. I don't remember exactly, but I think the outcome was that we would use a combination of articles and categories for the time being, but that new software should be written to enable a "hybrid" approach. Well... we're still waiting :-) Don't get me wrong, I recognize that most (all?) of the MediaWiki programmers are just hacking in their spare time, and aren't being paid, but Commons really needs such a system put in place.
Maybe that could be achieved through combining categroies, "image box" annotations, gallery inclusions, subcategories, and category intersections, to some abstract concept of annotation: * An annotation can concern only a part of the image (image box) * An annotation can concern the whole image
Both types of annotation use categories to store the information. Example: [1] is in both category "Bee" (through image box) and in gallery "Valeriana officinalis". The latter is in Category:Valerianaceae, which is a sub-sub-sub...category of Category:Plantae. So, all the information to find "Flower [plantae] and bee" is there. It would "only" (biiiig olny) need an algorithmical approach to combine them so [1] would show up in the results.
As for metadata (IPTC etc.) see my recent mail about storing template variables in their own searchable table. This could be combined into the annotation concept as well.
Cheers, Magnus
[1] http://commons.wikimedia.org/w/index.php?title=Image:Valeriana_officinalis_2...