[01:59:49] Alchimista [02:00:01] So you are trying to compare your tool to revision scoting? [02:00:06] *scoring [02:00:31] you could also take revision scoring scores as a feature [02:00:50] if you do not want to only rely on the scores that is [02:51:49] YuviPanda: hey, Around? [10:05:00] ToAruShiroiNeko: hi [10:06:12] right now i was interested on using revscoring to test my bot wich is in development, but for the future i'm planning to use a AI too, provably using a modified version of revscoring [14:21:23] FYI: https://github.com/wiki-ai/revscoring/issues/164 [16:01:19] Hey folks! [16:01:28] We're going to do skype today so that we can hear Amir1 [16:01:32] ToAruShiroiNeko, ^ [16:01:56] * YuviPanda waves groggilllyyy [16:02:02] Very very groggily [16:02:15] halfak: as an FYI I'm on vacation next week [16:02:32] Thanks for the reminder :) [16:02:46] I have a couple PRs I want to have you look at before you go. I'll send an email with links. [16:03:25] For some reason, Skype has the "call" button grayed out. [16:03:29] And Amir1 isn't online. [16:06:22] AAron I dont here you :p [16:06:23] Amir1, I am talking! [17:06:27] Victory [17:06:51] I should have fireworks to use at the end of each of our meetings :P [17:09:30] :)))) [17:09:58] let me fix the issue of bwds system [17:14:56] halfak: https://nl.wikipedia.org/wiki/Wikipedia:De_kroeg#Wiki_labels_.26_Revision_Scoring_as_a_Service_for_Dutch_Wikipedia [17:15:09] so I have been working on that text for several days with breaks. [17:15:25] Do you want to add to it or modify it? [17:15:35] Other wise it will be my template for other languages [17:31:15] Seems like we shouldn't be posting in major forums in English. [17:31:26] Maybe we should find a translator before we post the next one [17:32:03] ^ +1 [17:32:09] Dutch isn't so hard to find [17:32:14] * YuviPanda pointedly looks at valhallasw`cloud [17:32:22] * valhallasw`cloud runs around in circles [17:32:31] :) [17:33:05] valhallasw`cloud, any chance you would have time to help us get that message translated before someone gets upset at our bad behavior? [17:34:33] halfak: somewhere after dinner, I think, but you're probably fine for nlwiki [17:34:44] good to know :) [17:34:50] except a general thing that from the readover it's not clear to me what the message is [17:35:01] +1 [17:35:15] I also find it to be a bit long and not to the point [17:36:21] 1) we want to help you combat vandalism 2) we need human input to configure our algorithms 3) when this is succesful, vandalism edits will be tagged? [17:36:59] or is it just to provide an api local people could hook into? [17:37:15] We'll just provide an API, but there's already several tools that hook into it. [17:37:30] Tools that are designed to work cross-wiki. [17:37:37] also there's https://gerrit.wikimedia.org/r/#/q/project:mediawiki/extensions/ORES,n,z that legoktm started [17:37:56] :)))) [17:38:42] halfak: right, so the point is 'once this is done, this enables you to use the following tools to combat vandalism: a, b, c and d. We also provide an API for local vandalism patrol tools to hook into' ? [17:41:20] halfak: do you have a list of those tools? [17:41:37] https://meta.wikimedia.org/wiki/Research:Revision_scoring_as_a_service#Tools_that_use_ORES [17:41:40] valhallasw`cloud, ^ [17:42:32] * halfak adds in development extension to the list [17:45:23] halfak: ok, added dutch summary/explanatiion [17:45:29] Thank you! [17:49:33] Anyone who feels like digging into our dependency injection system and how we do error handling, check out https://github.com/wiki-ai/revscoring/pull/165 [17:53:15] DI? Such Enterprise software [18:08:36] I will add that info [18:21:59] White_Cat: where can WP:Labels be translated? [18:23:12] valhallasw`cloud it can be any page on dutch wikipedia [18:23:23] Project space may make more sense [18:24:36] eh [18:24:38] okay [18:24:42] I still don't get what has to happen [18:27:52] valhallasw`cloud: I think he meant the wp:labels page on wiki while you are asking about the software [18:28:14] valhallasw`cloud I can copy paste the structure and you can translate it if you like [18:28:44] halfak did we not get german bad words processed? [18:28:49] I think you were emailed that [18:28:55] YuviPanda: no, I meant the page. The original post said "A localization on en:Wikipedia:Labels so that we have a local landing page." [18:28:55] can you forward that email? [18:29:08] and my question was 'where/hlow should it be localized' [18:29:23] valhallasw`cloud first task is to find a translation to the word labels [18:29:33] when we say labels we mean like "Wikipedia labels foo" [18:29:42] where it is more of a verb [18:30:03] and when we mean labels it is more like a name tag, like your nametag labelling you on wikipedia [18:31:29] sorry, I don't get it [18:31:46] what is 'foo'. The label? the page? [18:33:56] halfak: ^ do you think you can clarify? [18:34:09] oh no [18:34:12] foo is the campaihnm [18:34:18] so Wikipedia Labels Edit Quality [18:34:39] where we ask the users to look at diffs and marke them in good/bad faith and productive/damaging [18:34:48] YuviPanda I think Aaron is Busyron :) [18:35:07] ABusyron? [18:35:08] :p [18:43:56] I'm trying to schedule a short meeting to chat about https://phabricator.wikimedia.org/T106838, but I can't see other people's calendars. Should we just figure out a good time via IRC? [18:45:06] What timezones are we? Looks like mornings are best? [18:45:24] err, morning Pacific [19:36:50] halfak: I scheduled a meeting with you in about... 2h to talk to Timo. hope that's enough notice :| sorry! [19:37:00] halfak: both me and him won't be in the same place anymore next week [19:37:43] Sounds good :) [19:38:15] halfak: ok [20:00:46] halfak can you forward me the polish badwirds email? [20:00:52] I vividly remember it was processed [20:01:06] likewise do we have one for russian? [20:01:09] or ukranianb? [20:02:17] ToAruShiroiNeko, I'm not finding an email for it [20:02:46] none of the three? [20:04:18] I had a random thought about badwords and friends... What would you think about onwiki configuration for those parameters? [20:04:33] hah! [20:04:37] one step ahead of ya :) [20:04:41] https://meta.wikimedia.org/wiki/Research:Revision_scoring_as_a_service/Word_lists [20:04:58] Woot! Now we're talkin' [20:05:13] each sub page has bad word list if they are available [20:05:28] Hell yeah. So are you planning to pull that stuff during model generation, or is this just for human reference at the moment? [20:05:37] a bit of both [20:05:44] the TF IDF will output to meta [20:05:53] and humans would review it [20:06:10] and we can machine read it off of it and import it to revision scoring [20:06:10] awight, we can't have the badwords list change between model generation and application [20:06:28] this is true as well [20:06:35] But we could store the lists in the model for future use, I guess. [20:06:38] pulling bad word list would be semi automatic. [20:07:55] halfak so we lost all of base's work? [20:07:58] in Ukranian? [20:08:11] ToAruShiroiNeko, I don't know if I ever received it. [20:08:18] Amir might have [20:08:22] since he was talking with him [20:08:29] Amir1 :) [20:08:58] let me chekc [20:09:14] but there was another user interested to do the job [20:09:22] we can ask them [20:12:02] nope I didn't receive anything [20:36:37] ToAruShiroiNeko, how do you know that we got this word list in the first place? [21:02:59] https://meta.wikimedia.org/w/index.php?title=Research:Revision_scoring_as_a_service/Word_lists/sco&diff=13037226&oldid=13034290 [21:03:09] what do you think of this halfak ? [21:04:00] Oh! It was the bot that make the post! [21:04:11] I was thinking, "Yes. That looks like a wordlist." [21:04:30] and stop words are there too [21:04:31] Once suggestion: Sort it alphabetically. [21:04:35] The scores don't really matter. [21:04:38] both bad words and stop words [21:05:18] it goes for human review and it may help them (I can say it's improbable but it's not impossible) [21:05:53] Amir1, when I work with these lists, the first thing I do is strip off the score and sort alpha. I suppose others might not. [21:06:10] okay [21:06:20] gediz is saying the same, I'm down for it [21:06:39] Secret gediz messages [21:06:54] * halfak squints and looks side-to-side [21:33:43] https://meta.wikimedia.org/wiki/Research:Revision_scoring_as_a_service/Word_lists/sco [21:33:48] what about now? [21:34:12] Awesome! [21:34:23] "Adverb" O.O [21:34:37] it's a very small wiki [21:34:43] I don't except more [21:34:46] "Aboot" :D [21:34:56] :D [21:37:28] I just updated the code in github [21:37:28] https://github.com/Ladsgroup/Bad-Words-Detection-System [21:50:32] BWDS [21:50:51] Aboooooooooot [21:52:05] I couldn't find the correct way to make BDSM [21:52:08] :D [21:53:49] Badword Detection System for Men? [21:56:38] Badword Detection Supervised Machine learning [21:56:53] Engrish but works [21:58:17] :)))))