Il 03/06/2010 22:26, Hanno Schlichting ha scritto:
> On Thu, Jun 3, 2010 at 9:50 PM, Duco Dokter
> <dokter@...> wrote:
>> Okido, I'll ponder over the weekend and a couple of beers. Maybe I'll start
>> working on some add on. In Holland nowadays the government is big into
>> metadata, and has defined huge sets to be used with content. I reckon things
>> will be the same elsewhere, so a good metadata solution may come handy.
> Metadata is in a weird state, from what I can tell, my oversimplified
> take on it (when it comes to general web content):
> - In the early web days people thought metadata is cool, things like
> Dublin Core emerged and were used quite a lot
> - Search engines started taking it into account
> - People started abusing metadata to get better search rankings
> - Search engines stopped looking at metadata at all, instead they
> looked only at the user-visible content of a page
> - People started talking about semantic web, ontologies, taxonomies
> etc. which still remain irrelevant for general content, there's
> applications in special purpose and academic settings
> Today most freely editable metadata is irrelevant for most purposes.
> Search engines focus on better information retrieval and try to
> distill information from the actual content. Other projects try to do
> automatic classification of content. Google Mail and the no-folders
> approach is one such example of the modern "search based" information
> There's a different crowd that likes manual detailed classification of
> content. Controlled vocabularies, strict policies on naming, 10 folder
> levels deep hierarchies and the like. For specific use-cases this
> makes sense, like tagging news items (politics, international,
> climate, ...) but for a general web page it's hard to see how this can
> add much value. There's a large time overhead for doing it and
> unsolved problems around getting a consistent classification from a
> diverse group of content editors.
> Some governments fall into the second crowd, but certainly not all do.
> Some governments have adopted policies that reflect the state of the
> web 10 years back, as their procedures cannot keep up with the
> changing ecosystem. We have seen the same in accessibility guidelines
> that require XHTML strict for no good reason.
> There's a subset of metadata that is very much useful and used, like
> modification times, language, responsible person, copyright
> information, etc. But the freely editable "keywords" kind seems
> dubious to me. I've seen lots of tag-clouds or other representations
> of this type of information, but I haven't seen anyone, who showed
> that these are more than visual distractions.
Working in the a librarian system, I can agree with Hanno. Metadata &
C. are good for projects (Europe will ask you a lot ;) ), for strict
classification and semantic of a document.
For web content, rather than metadata, we need to move to Information
Retrieval technologies. For tagging with keywords, it is useful mostly
for using Collections in Plone and grouping content, so Plone does well
Said that, a lot of customer of Plone in Italy are... public
institutions, so paid a little attention to metadata handling is good.
It can be cosmetic ( a better interface to categories/keyword handling),
or substantial (ontologies/thesaurus for dummies).