Ettore,
Wikidata has the ability of crowdsourcing...unfortunately, it is not
effectively utilized.
Its because Wikidata does not yet provide a voting feature on
statements...where as the vote gets higher...more resistance to change the
statement is required.
But that breaks the notion of a "wiki" for some folks.
And there we circle back to Gerard's age old question of ... should
Wikidata really be considered a wiki at all for the benefit of society ?
or should it apply voting/resistance to keep it tidy, factual and less
messy.
We have the technology to implement voting/resistance on statements. I
personally would utilize that feature and many others probably would as
well. Crowdsourcing the low voted facts back to applications like
OpenRefine, or the recently sent out Survey vote mechanism for spam
analysis on the low voted statements could highlight where things are
untidy and implement vote casting to clean them up.
"...the burden of proof has to be placed on authority, and it should be
dismantled if that burden cannot be met..."
-Thad
+ThadGuidry <https://plus.google.com/+ThadGuidry>
On Sat, Sep 29, 2018 at 2:49 AM Ettore RIZZA <ettorerizza(a)gmail.com> wrote:
Hi,
The Wikidata's ontology is a mess, and I do not see how it could be
otherwise. While the creation of new properties is controlled, any fool can
decide that a woman <https://www.wikidata.org/wiki/Q467>is no longer a
human or is part of family. Maybe I'm a fool too? I wanted to remove the
claim that a ship <https://www.wikidata.org/wiki/Q11446> is an instance
of "ship type" because it produces weird circular inferences in my
application; but maybe that makes sense to someone else.
There will never be a universal ontology on which everyone agrees. I
wonder (sorry to think aloud) if Wikidata should not rather facilitate the
use of external classifications. Many external ids are knowledge
organization systems (ontologies, thesauri, classifications ...) I dream of
a simple query that could search, in Wikidata, "all elements of the same
class as 'poodle' according to the classification of imagenet
<http://imagenet.stanford.edu/synset?wnid=n02113335>.