Hello Tom,
Is there a description somewhere of what each state
means?
Here is the meaning of each state:
* approved: statements approved with the UI and added to Wikidata
* unapproved: basic state for statement added to the database. They are the only ones that
are presented to the user
* wrong: the statement have been rejected throw the UI (using both claim and reference
"reject" buttons)
* skipped: statements that have been loaded into the database but hidden for some reason
(it is the case of census data of U.S. cities)
* othersource: currently unused I think
* duplicate: detected as duplicate by PrimarySources. It may happen if the statement have
been created in Wikidata after its upload to Primary Sources
* blacklisted: the item is blacklisted for some reason like a reference URI filtered:
https://www.wikidata.org/wiki/Wikidata:Primary_sources_tool/URL_blacklist
If so, it it seems like a bunch of information that
would be useful for analysis is being lost.
I agree. But we should find a tradeoff
between lost of information and simplicity of the UI (I am not sure that 5 slightly
different reject buttons will be very user friendly).
Cheers,
Thomas
Le 1 oct. 2015 à 18:27, Tom Morris
<tfmorris(a)gmail.com> a écrit :
The version with support for all states has apparently been deployed. The default is
"unapproved" and "any" is still the catchall, but you can also use:
approved, wrong, skipped, othersource, duplicate, blacklisted
e.g.
http://tools.wmflabs.org/wikidata-primary-sources/statements/all?state=dupl…
Is there a description somewhere of what each state means? I assume that things that are
marked as "blacklisted" never got presented to the user. Is that true and is it
also true of "duplicate"? If so, these are really tracking quality of the data
prep process (ie this filtering should have been done before the data was loaded into the
backend).
In looking at the UI, I only see Accept & Reject buttons. Does this mean that Reject
(ie state == wrong) is the union of duplicate fact, bad reference, we've got enough
references so I don't feel like adding another one, bad fact, and any other reason for
rejection? If so, it it seems like a bunch of information that would be useful for
analysis is being lost.
Tom
On Tue, Sep 29, 2015 at 6:10 PM, Tom Morris <tfmorris(a)gmail.com> wrote:
Thanks for the quick reply, Thomas -- and for creating the issue.
On Tue, Sep 29, 2015 at 4:15 PM, Thomas Tanon <thomaspelta(a)gmail.com> wrote:
Sadly, the "state" parameter currently only accepts the values
"unapproved" and "all".
When "all" didn't work, I had a peek at the sources and discovered that the
only legal value for the state parameter is actually "any".
The following query works for the analysis that I want to do (albeit with a *lot* of
filtering required):
http://tools.wmflabs.org/wikidata-primary-sources/statements/all?state=any
Tom