Talk:Global AbuseFilter/Archives/2024

Abuse filter 110: Adding emoji unicode characters

Hi, the emoji range 🌀-🙏 in Special:AbuseFilter/110 is too wide as it will match to Extended Unicode Characters (special characters, diacritics, or extended Latin characters) which are in the range. ( example log entry , example log entry2)

One can test matching also with following command.

echo "Tadžikistanin" | grep -E "[🌀-🙏]"

One solution would start to filter out incorrect blocks at the end part. Following rule would allow "ž" style characters so it could be added to the rule:

!added_lines irlike "[\p{M}\p{Latin}]" &

ie updated filter would look like this:

!"autoconfirmed" in user_rights &
page_namespace == 0 &
(
   emoji := "[🌀-🙏🚀-🛼🤌-🧿🩰-🫶☀-☄☇-♬♰-✒✙-➿۩]";
   rcount(emoji, added_lines) > 0 &
   (
       rcount(emoji, added_lines) > 
       rcount(emoji, removed_lines)
   ) & (
       page_title != "Emoji" &
       !added_lines irlike "[\p{M}\p{Latin}]" &
       !added_lines irlike "{{unicode\|" &
       !added_lines irlike "{{nihongo\|" &
       !new_wikitext irlike "\[\[Category:Unicode blocks\]\]" &
       !new_wikitext irlike "Esta nota contiene algunas 📆instrucciones" &
       !equals_to_any(wiki_name, "frwiki", "nlwiki")
   ) 
)

--Zache (talk) 08:36, 19 July 2024 (UTC)

@Zache a side effect of your proposed change would be, that Special:AbuseFilter/110 no longer logs many edits where emojis are actual added. If you try your proposed change on fi:Toiminnot:Väärinkäyttösuodatin/examine/log/924692 or fi:Toiminnot:Väärinkäyttösuodatin/examine/log/924112 both edits wouldn't be caught by the filter.

Actually after checking log entries of different projects, most emoji additions globally would no longer be logged with your proposed change. Given that the filter is just logging and not taking any action, I think it's better to keep it that way.

There appears to be a fiwiki version of the global emoji filter which disallow edits: fi:Toiminnot:Väärinkäyttösuodatin/152. Again with your proposed update it would prevent almost no emoji additions at all, but that's the filter where changes are more useful, because false positives in that filter actually do some damage by preventing potentially useful edits...

en:Special:AbuseFilter/680 appears to be more advanced, you might want to check if those rules are better in preventing false positives while still catching actual emoji additions. Johannnes89 (talk) 11:07, 19 July 2024 (UTC)

Yeah, and thanks, I will update the fi:Toiminnot:Väärinkäyttösuodatin/152 --Zache (talk) 12:41, 19 July 2024 (UTC)