I followed these steps a few moments ago (using Brave browser on Ubuntu):

1. Go to https://pt.wikipedia.org/wiki/Categoria:Teoria_dos_n%C3%B3s?uselang=en&skin=vector
2. Click on Add Links on the Languages section of the sidebar
3. Copy/paste "enwiki" to the "Language" field
4. Copy/paste "Category:Knot theory" to the Page field
5. Press ENTER and wait for the "Link with page" confirmation screen
6. Press ENTER again to save

Nope, so I went ahead and just disabled the rules.

Thank you! This worked:

const p = mw.util.addPortlet( 'foo', 'Foo portlet', '#p-interaction' );
mw.util.addPortletLink('foo', '#', 'New portlet link');
mw.util.addPortletLink('foo', '#', 'New portlet link 2');
if ( p ) { p.parentNode.appendChild(p); }

That is great! Thank you! 😃

See also: T303488: [Gadget support] Add a Javascript function to create a new portlet section similar to mw.util.addPortletLink

In T151008#9079802, @waldyrious wrote:

From the counts it seems like the singular zero has more hits than plural zero. Does the "distribution of person and number" from the previous comment contradict this? I couldn't understand the output format used there.

I believe the initial plan was to first finish the task T113682: Finish conversion from LQT to Flow on pt.wikibooks and then that would allow archiving the old LiquidThreads topics by leaving them read-only (current task).

Having said all that, the ideal solution would be for abuse filters to provide more contextual feedback (T174554) and allow to define some rules as being safe to be checked in real time. this would allow all kinds of editors (not only Content Translation) to provide early and contextual guidance to users.

What is blocking this?

I'm unable to reproduce this at the moment (tried on Firefox and Brave browsers)

I'm also seeing translated content disappear when it is inside templates. E.g.: the paragraph inside "Citation needed span" in
https://en.wikipedia.org/wiki/Quadratic_voting?oldid=1012353281
Even if I make some edit after that paragraph, and wait for it to be saved, the English content is restored and the translation is lost.

I'm ok with closing it as declined.

In T269632#6674823, @bd808 wrote:

It looks like there are quite a few other Content-Security-Policy violations for this tool: https://csp-report.toolforge.org/search?ft=ptwikis

I believe fixing T134678: ContentTranslation should generate an AbuseFilter log whenever it shows a warning for the users would help debugging this.

This happened in most/all of my recent translations using the tool, and it also happens for the translation which is already in progress, at
https://pt.wikipedia.org/wiki/Special:ContentTranslation?page=Group+extension&from=en&to=pt&targettitle=Extens%C3%A3o+de+grupo

I had to use the same workaround again for https://pt.wikipedia.org/w/index.php?diff=60053113

For some reason I get 404 Not Found from https://cxserver.wikimedia.org/v2/page/en/Word2vec while https://cxserver.wikimedia.org/v1/page/en/Word2vec seems to work fine

Almost three years later, I'm still unable to start a new translation of https://en.wikipedia.org/wiki/Word2vec
Now it says:

Loading the saved translation...

followed by

The "Word2vec" page could not be found in English Wikipedia

which is incorrect, given the link above is working.

Still happening: https://pt.wikipedia.org/wiki/Simetria_de_reflex%C3%A3o?diff=59627837#footer

@Danilo made these tools for that:
https://ptwikis.toolforge.org/FiltroIP
https://ptwikis.toolforge.org/Filtros:180

In T264940#6534046, @Danilo wrote:

Number of edits: "Edits per day" graph tool. I have created that tool with a query similar to that I used to get the active users. The graph show us that IPs used to make approximately 1700 edits per day. After the mandatory registration the new users edits have raised approximately 700 daily edits (from ~700 to ~1400), that suggest that about 700 edits that was made by IPs become to be made by new registered users and about 1000 are no longer been made.
(...)
Quality of edits with ORES: https://quarry.wmflabs.org/query/48860. I used the ORES damaging model to estimate the proportion of damaging edits. The data shows that it has decreased from approx. 18% to approx. 7%. That suggest us that those approx. 1000 edits per day that are no longer been made by IPs are worse edits then those approx. 700 that become to be made by new registered users.

In T264622#6517839, @Daimona wrote:

There you go, the stats were reset and the filter was throttled. Likely some caching issue. T264629 could help, probably.

Indeed, now the main page says it is "Enabled, throttled":
https://pt.wikipedia.org/wiki/Special:AbuseFilter?offset=179&limit=1&uselang=en

I wonder if this deserves a higher priority now, given that https://gerrit.wikimedia.org/r/609773 implemented some kind of metadata (T254074: Implement the reverted edit tag). This feature should help with part of the concerns raised at

In T261133#6419171, @Darwinius wrote:

...
As for Huggle, not only there is a chronic problem of people willing to waste their volunteer time there operating it, instead of creating content, but it's a bad solution as well, since it fills up the historic of the articles with revertions after revertions, polluting it and making it much less readable.
...

@Danilo generated the following table comparing articlequality scores for the latest version of all articles to the scores which would be produced by the Python script which is/was used to make bot assessments:

MariaDB [s51206__ptwikis]> SELECT pe_qualidade, SUM(pe_qores = 0) ORES_0, SUM(pe_qores = 1) ORES_1, SUM(pe_qores = 2) ORES_2, SUM(pe_qores = 3) ORES_3, SUM(pe_qores = 4) ORES_4, SUM(pe_qores = 5) ORES_5, SUM(pe_qores = 6) ORES_6 FROM page_extra GROUP BY pe_qualidade ORDER BY pe_qualidade;
+--------------+--------+--------+--------+--------+--------+--------+--------+
| pe_qualidade | ORES_0 | ORES_1 | ORES_2 | ORES_3 | ORES_4 | ORES_5 | ORES_6 |
+--------------+--------+--------+--------+--------+--------+--------+--------+
|            0 |  68218 |      0 |      0 |      0 |      0 |      0 |      0 |
|            1 |      3 | 618819 | 204187 |  27523 |   1847 |   3562 |   3261 |
|            2 |      0 |   5565 |  69323 |  24777 |   1390 |   7496 |    350 |
|            3 |      0 |     71 |    472 |  14361 |   1861 |   7412 |    572 |
|            4 |      0 |      5 |     10 |   2948 |   2361 |   2978 |   1208 |
|            5 |      0 |      0 |     16 |     59 |    136 |   1056 |    161 |
|            6 |      0 |      0 |      0 |     35 |    190 |    188 |    782 |
+--------------+--------+--------+--------+--------+--------+--------+--------+
7 rows in set (3.70 sec)

(the label is set to zero if the quality is unknown, possibly due to the page being deleted)

See also: https://en.wikipedia.org/wiki/User:Joeytje50/JWB

Possibly related to https://github.com/azatoth/twinkle/commit/8f7b2f367276c6cf8e0ef78b82d9957415221780

I confirmed this by replacing my global.js by console.log( 'Started global.js.' ); and then loading
https://pt.wikipedia.org/wiki/Special:BlankPage?debug=true
There should be a log in the console, but it was not there.

Progress (100% done, -19 labels left):

I've submitted https://github.com/wikimedia/articlequality/pull/132

Updated info (as of commit c3a66b0 plus the specific changes which define each of the tests):

• Full period; Bots removed

accuracy (micro=0.8, macro=0.861):
	    1      2      3      4      5      6
	-----  -----  -----  -----  -----  -----
	0.781  0.827  0.877  0.899  0.875  0.908

$ cat datasets/ptwiki.labelings.20200301.remove_bots.json | json2tsv wp10 | sort | uniq -c
 145657 1
  32807 2
   6177 3
   2346 4
   1646 5
   1542 6
$ cat datasets/ptwiki.balanced_labelings.9k_2020.remove_bots.json | json2tsv wp10 | sort | uniq -c
   1500 1
   1500 2
   1500 3
   1500 4
   1500 5
   1328 6

• Since 2014 (inclusive); Bots included

accuracy (micro=0.81, macro=0.875):
	    1      2      3      4      5      6
	-----  -----  -----  -----  -----  -----
	0.799  0.806  0.867  0.915  0.928  0.933

$ cat datasets/ptwiki.labelings.20200301.since_2014.json | json2tsv wp10 | sort | uniq -c
   7537 1
   3346 2
   1276 3
    690 4
    653 5
    684 6
$ cat datasets/ptwiki.balanced_labelings.9k_2020.since_2014.json | json2tsv wp10 | sort | uniq -c
   1500 1
   1500 2
   1276 3
    690 4
    653 5
    684 6

@Halfak: Oops... I missed the -v flag when I used grep to remove the bot assessments. So, instead of considering only human assessments, I extracted only the bot assessments! Once I add that flag, the number of assessments by humans seems more reasonable:

$ cat datasets/ptwiki.labelings.20200301.user.json |grep -v -P '"user": "[^"]*([Bb][Oo][Tt]|[Rr][Oo][Bb][ÔôOo])[^"]*"' | json2tsv wp10 | sort | uniq -c
  28403 1
  13343 2
   5329 3
   2209 4
   1458 5
   1281 6

In this case, the explanation for such a high accuracy is likely that the bots assessments are very predictable (it is hardcoded in their dna code ;-).

For future reference: there is now a translation at https://pt.wikipedia.org/wiki/User:EpochFail/ArticleQuality

@GoEThe: in case you have any suggestions on better images for this purpose, we can try changing them. @Halfak suggested the https://commons.wikimedia.org/wiki/Category:OOUI_icons as a good source of icons we could use.

@GoEThe : I see you've installed the version of the script I mentioned at T246667#6079484. Did you have the chance to test it on Special:Newpages? Is it good enough for us to publicize it for other users?

PS: I didn't change the thresholds in the Makefile, so the samples were not as balanced as might be wanted:
(Note: By mistake, I forgot the -v flag in the grep above, so the results for the first case are inverted, that is, they contain bot_only, instead of no_bots)

$ cat datasets/ptwiki.balanced_labelings.9k_2020.no_bots.json | json2tsv wp10 | sort | uniq -c
   1500 1
   1500 2
    759 3
     20 4
     95 5
    203 6
$ cat datasets/ptwiki.balanced_labelings.9k_2020.since_2014.json | json2tsv wp10 | sort | uniq -c
   1500 1
   1500 2
   1247 3
    674 4
    630 5
    654 6

In T250809#6122588, @Halfak wrote:

Wow. This is really awesome. I wonder what would happen if we retrained the models on recent data only. In enwiki we found that the definition of quality changed over time. There should be plenty of observation after 2014 to give us good signal.

Halfak fixed this in https://github.com/wikimedia/wikilabels/pull/263

See https://github.com/wikimedia/wikilabels/pull/264

I've tried to do this:

diff --git a/Dockerfile b/Dockerfile
index 5fc2d0d..19f39a5 100644
--- a/Dockerfile
+++ b/Dockerfile
@@ -5,7 +5,8 @@ RUN apt-get update && apt-get install -y \
     g++ \
     python3-dev \
     libmemcached-dev \
-    libz-dev
+    libz-dev \
+    memcached

After patching¹ the extractor to also collect user names, I found that these are the top 10 users who added/modified the most assessments:

FMTbot            91366
Rei-bot           25155
BotStats          15830
Fabiano Tatsch    14829
Leandro Drudo      4660
GoEThe             3172
Burmeister         3128
Rei-artur          2965
FilRBot            2444
VítoR Valente      1895

Then I produced² the following graphs showing the number of labels added/modified by bots³ by year, for each of the six quality levels. There are many quality 1 and 2 assessments made by bots.

Here are some graphs showing the evolution of the assessments extracted from ptwiki:

It occurred to me that some of these expressions are also used by Salebot¹, with the difference that in the bot config² users assign a score to each word/regex indicating how much it contributes towards classifying an edit as needing to be reverted. This allows it to "ignore" words which are common in good edits, unless there are too many of them.

See https://github.com/wikimedia/articlequality/pull/127 for a possible solution.

It is not uncommon for some good faith edit to add a new expression (or badly written regex) to such lists and then breaking (to some extent) the tools which use them (e.g. increasing its false positives).

Here are some examples of existing lists, of varying quality and formats, used by other tools:

• https://pt.wikipedia.org/wiki/WP:Software/Anti-vandal_tool/badwords
• https://pt.wikipedia.org/wiki/User:Alchimista/Aleph_Bot/Express%C3%B5es
• https://pt.wikipedia.org/wiki/User:Salebot/Config
• https://pt.wikipedia.org/wiki/WP:Projetos/AntiVandalismo/Express%C3%B5es_problem%C3%A1ticas
• https://pt.wikipedia.org/wiki/User:O_Politizador/curiosidades_apagadas/Lista_de_express%C3%B5es_idiom%C3%A1ticas_com_palavras_de_baixo_cal%C3%A3o

I've updated the patch.

Is this still wanted nowadays?

While the dumps are processed, we could store the <id> of the talk pages instead of their <title>s. Then, an API query such as
https://pt.wikipedia.org/w/api.php?action=query&format=json&prop=info&pageids=18363&formatversion=2&inprop=subjectid
will return the <id> of the associated subject page (the one whose text we are interested in). This should work when pages are moved, since page moves do not change the pageid (but it is not guaranteed if the page is deleted and restored).

Could the number of labels per article have a negative impact on the quality of the model?
These are the frequency of the number of labels/page in the full set and in the 9k sample:

$ cat ptwiki.labelings.20200301.json | json2tsv page_title | sort | uniq -c | cut -c-8 | sort |uniq -c
 181477       1 
   3042       2 
    517       3 
    100       4 
     19       5 
      2       6 
      2       7 

The following pull request is related to improving the articlequality model: https://github.com/wikimedia/articlequality/pull/122

See https://github.com/wikimedia/articlequality/pull/122 for another possible explanation for the problem:

In T251171#6087118, @Halfak wrote:

It does include the new articles matched beyond the ER# tags. Could it be possible that we're not matching the features effectively? Maybe we could generate a sample of articles and the values of the features the spam and vandalism articles. Maybe there's a bug in the extraction that is hard to see.

I didn't train extract/retrain the model after the change to verify its impact on the metrics, but I believe it might help by improving the dataset quality.

@Halfak do you have a quick way to get how many assessments by each user in the dataset ptwiki.balanced_labelings.*_2020.json which was used for articlequality model? Are we getting labels from a diverse set of users or mostly from just a few users?

It could be. For example, @Darwinius noticed that images loaded from Wikidata are not counted:
https://www.mediawiki.org/wiki/ORES/Issues/Article_quality?diff=3804470
I wouldn't be surprised if there was some obscure problem with feature extraction.

@GoEThe Correct me if I'm mistaken, but I believe a reasonable amount of new articles having vandalism or spam would contain expressions such as the words_to_watch mentioned by Halfak. For reference, the expressions are listed at
https://github.com/wikimedia/revscoring/blob/76c737f2998bbba5b5dd942823f43383f1a4b47e/revscoring/languages/portuguese.py#L153-L189

That is odd. Does this tuning report reflect only the changes in the ptwiki features, or does it also include other articles to the dataset as mentioned at T246667#6067366?

@Halfak: what should be considered as a true positive in these multi-class classification problems? (when filling the template misclassification report at mw:ORES/Issues/Article quality)
Would a "featured article" be a "positive" case for the articlequality model, and any other level is a "negative"? Or something else?
What about the draftquality model? (in this case it does not even seem to have any implicit order between the classes (e.g. nothing like OK < spam < unsuitable)

In T250809#6078671, @Chtnnh wrote:

@He7d3r has updated the message. What do you think about it now? I think we should see some input from the community on the model (articlequality) soon. In the meantime what can we achieve for draftquality?