Research talk:Revision scoring as a service/Work log/2015-11-12

Thursday, November 12, 2015 edit

Today, I'm loading Russian and Ukranian edit quality campaigns into Wiki labels.

u_wikilabels=> SELECT id, name, wiki FROM campaign WHERE wiki IN ('ruwiki', 'ukwiki');
 id |                         name                         |  wiki  
----+------------------------------------------------------+--------
 10 | Редактировать качество (20k случайная выборка, 2015) | ruwiki
 11 | Редагувати якість (20k випадкова вибірка, 2015)      | ukwiki

OK. Campaign records created. I'm using the result of running the prelabeling script on these wikis to load them up.

First, let's look at some sample output:

$ cat datasets/ruwiki.prelabeled_revisions.20k_2015.tsv | grep -P "rev_id|[0-9]+\tTrue" | head | /srv/wikilabels/venv/bin/wikilabels task_inserts 5 
INSERT INTO task (campaign_id, data) VALUES
  (5, '{"reason": "NULL", "needs_review": "True", "rev_id": 66696639}'),
  (5, '{"reason": "reverted edit", "needs_review": "True", "rev_id": 66852339}'),
  (5, '{"reason": "NULL", "needs_review": "True", "rev_id": 71887614}'),
  (5, '{"reason": "NULL", "needs_review": "True", "rev_id": 70870700}'),
  (5, '{"reason": "NULL", "needs_review": "True", "rev_id": 71123011}'),
  (5, '{"reason": "reverted edit", "needs_review": "True", "rev_id": 69314507}'),
  (5, '{"reason": "NULL", "needs_review": "True", "rev_id": 72741611}'),
  (5, '{"reason": "NULL", "needs_review": "True", "rev_id": 73622657}'),
  (5, '{"reason": "NULL", "needs_review": "True", "rev_id": 73276754}');

Now to run the real thing.

halfak@wikilabels-01:~$ cat datasets/ruwiki.prelabeled_revisions.20k_2015.tsv | grep -P "rev_id|[0-9]+\tTrue" | /srv/wikilabels/venv/bin/wikilabels task_inserts 10 | psql -h wikilabels-database --user u_wikilabels u_wikilabels -W
Password for user u_wikilabels: 
INSERT 0 5193
halfak@wikilabels-01:~$ cat datasets/ukwiki.prelabeled_revisions.20k_2015.tsv | grep -P "rev_id|[0-9]+\tTrue" | /srv/wikilabels/venv/bin/wikilabels task_inserts 11 | psql -h wikilabels-database --user u_wikilabels u_wikilabels -W
Password for user u_wikilabels: 
INSERT 0 3318

OK. Now to test on those wikis! ... Looks like uk:Вікіпедія:Оцінки and ru:Википедия:Оценка are working.  :) --Halfak (WMF) (talk) 18:11, 12 November 2015 (UTC)Reply


Looks like I'm doing idwiki, nlwiki and eswiki too.

u_wikilabels=> SELECT id, name, wiki, active FROM campaign WHERE wiki IN ('nlwiki', 'idwiki', 'eswiki'); id |                     name                     |  wiki  | active 
----+----------------------------------------------+--------+--------
 12 | Editar calidad (20k muestra aleatoria, 2015) | eswiki | t
 13 | Mengedit kualitas (20k sampel acak, 2015)    | idwiki | t
 14 | Kwaliteit bewerken (20k steekproef, 2015)    | nlwiki | t
(3 rows)
halfak@wikilabels-01:~$ cat datasets/eswiki.prelabeled_revisions.20k_2015.tsv | grep -P "rev_id|[0-9]+\tTrue" | /srv/wikilabels/venv/bin/wikilabels task_inserts 12 | psql -h wikilabels-database --user u_wikilabels u_wikilabels -W
Password for user u_wikilabels: 
INSERT 0 8434
halfak@wikilabels-01:~$ cat datasets/idwiki.prelabeled_revisions.20k_2015.tsv | grep -P "rev_id|[0-9]+\tTrue" | /srv/wikilabels/venv/bin/wikilabels task_inserts 13 | psql -h wikilabels-database --user u_wikilabels u_wikilabels -W
Password for user u_wikilabels: 
INSERT 0 2200
halfak@wikilabels-01:~$ cat datasets/dewiki.prelabeled_revisions.20k_2015.tsv | grep -P "rev_id|[0-9]+\tTrue" | /srv/wikilabels/venv/bin/wikilabels task_inserts 14 | psql -h wikilabels-database --user u_wikilabels u_wikilabels -W
Password for user u_wikilabels: 
INSERT 0 4177

--Halfak (WMF) (talk) 20:10, 12 November 2015 (UTC)Reply

Return to "Revision scoring as a service/Work log/2015-11-12" page.