[00:12:28] so many 1:1s [16:04:58] o/ [17:29:33] \o [19:34:39] J-Mo, I have extracted 446k Teahouse-related link insertions from the history of English Wikipedia. [19:34:56] well that's exciting, halfak. [19:35:11] Regretfully, it looks like some of those are insertions of [[WP:TEA]] [19:35:15] So I have to do some filtering :/ [19:35:29] [[WP:TEA?]] was what I was trying to match. [19:35:32] Yay. [19:35:33] should we plan to put a graph of them in the metrics deck? [19:35:37] yes [19:35:43] most excellent. [19:35:50] I'm hoping to get this graph read by next week. [19:36:25] I'm s'posed to have a draft of the presentation to Praveena by Thursday, but we can just put a GRAPH GOES HERE placeholder in the slides. [19:37:25] planning to exclude HostBot invites? since those are more or less constant, we might see trends easier if they weren't part of the set. [19:37:25] Yikes! Yes. [19:37:34] Thursday ._. [19:37:52] But Thanksgiving! [19:37:57] should be fine; I'm working on the deck now. Will share with you by EOD and we can talk it through tomorrow. [19:38:03] kk [19:38:24] you already had your Thanksgiving, guillom. Back to work! [19:38:59] guillom: the Teahouse is one thing I am very thankful for, so I don't mind working on it during holidays. [19:39:12] Emufarmers: I don't really care about Thanksgiving, but I will enjoy the holidays :) [19:40:02] (And I'll use them to read semi-work-related books.) [19:42:01] guillom: whatcha readin'? [19:43:03] J-Mo: https://office.wikimedia.org/wiki/User:Guillaume/Reading_log [19:45:24] interesting, guillom. I hadn't seen this lit review before. Totally awesome! [19:46:05] J-Mo: I'm mostly in the "absorbing & digesting" stage now. [19:46:16] I have a lot to catch up on :) [19:47:15] we all do, guillom. that's why I'm so thankful for masssly (and others') work on the research newsletter; it helps me stay kinda-sorta up to date. [19:47:43] +1 same here. Hard to build stuff, analyze stuff and still have time to be a good reader. [19:48:05] Agreed; My goal is to provide a topic-based summary, to complement the time-based newsletter. [20:15:30] halfak: re: teahouse, we now have direct (will go away at some point, unless we can get more resources!) elasticsearch access from labs [20:15:49] I can show you how to use it at some point if still desired, or you can point me to people who are [20:16:06] YuviPanda, interested, but not mental space ATM :( [20:16:21] halfak: that's ok! do you have vague people pointers? [20:16:48] (is ok if you don't!) [20:16:50] J-Mo would be likely to have some. [20:17:09] It would be great if we could get some design time for how we'd have people accessing elastic search for teahouse stuff [20:17:23] E.g. maybe a query that drops down while a user types of Q. [20:17:33] * YuviPanda nods [20:17:35] indeed. I'd love to test elasticsearch on the Teahouse question archive. [20:17:39] I've also never used the teahouse so not sure at all [20:18:07] I'm focused on the metrics prez today, but would be happy to meet and brainstorm next week sometime? [20:18:13] J-Mo: sure! [20:18:17] should I set up a meeting YuviPanda? [20:18:20] J-Mo: do you want to setup a meeting? [20:18:22] hah :D [20:18:24] yes [20:18:36] yes! Invite inbound… feel free to move it if you need to [20:19:07] J-Mo: \o/ <3 [20:19:09] ok [20:19:17] halfak: ty for people pointer :) [20:19:28] :D [20:20:02] sent. [20:20:14] added you as optional, halfak, in case you want to play with us [20:20:22] i've accepted, J-Mo [20:20:30] excellent!!! [20:26:26] elastic search for teahouse???? [20:34:15] hi harej [20:34:23] it's yuvipanda [20:34:24] we've an experimental elasticsearch replica in labs [20:34:28] and are looking for uses [20:34:30] what, exactly, is elasticsearch [20:34:39] how does it differ from run of the mill wikipedia search [20:34:46] so [20:34:52] run of the mill wikipedia search [20:34:57] exposes a subset of elasticsearch [20:35:02] it's like using mw APIs [20:35:04] vs the db replica [20:35:08] elasticsearch is the db replica [20:35:12] wikipedia search is our API [20:35:21] and exposing elasticsearch directly means you can do more powerful querying [20:35:23] like https://www.elastic.co/guide/en/elasticsearch/reference/1.4/_introducing_the_query_language.html [20:35:40] whooaaaaa [20:36:04] so this is kindof like db replicas but for current text of pages (no historical / deleted pages) [20:37:01] it currently has text from enwp, dewp, wikidata and commons [20:37:14] so, understanding that i have access to a more powerful search engine, what could I do with it that I can't do currently? [20:37:20] search full text, I suppose [20:37:21] basically [20:37:23] yeah [20:37:58] Most of what I work with involves article metadata, i.e. what's in the db replicas. [20:38:08] i think wikimedia ai people would be very interested [20:56:06] harej: yeah, probably. although I note that you can search for things like WPBannerMeta or discussions or what not but you can probably already do t hat [21:02:38] Well for our WikiProject/pagename index we use categories, and for new discussions we use edit summaries. [21:41:17] J-Mo, https://commons.wikimedia.org/wiki/File:Teahouse_links_additions.monthly.svg [21:41:49] I'm guessing that Feb 2012 was the Teahouse inception [21:42:05] It looks like most of the redirects are unused. [21:42:21] yep. February 27th [21:42:27] awesome [21:42:47] possible to sum them and publish that plot for the metrics deck? [21:43:03] oh, is there going to be a metrics meeting with teahouse stuff? [21:43:06] Want all of them summed? [21:43:50] yes please! 2012-present [21:44:27] YuviPanda: yep, next metrics meeting. we have 10 glorious minutes to explain what the Teahouse is, what we found, why it matters, and what should be done next… [21:44:53] nice! [21:45:47] this is a great piece of data, halfak. Shows how effectively the Teahouse has been integrated into wiki culture. [21:46:37] J-Mo it's a little weird to sum these up. [21:46:51] why? they're all links to the Teahouse [21:46:57] Since a posting that includes [[Wikipedia:Teahouse]] might also include [[Wikipedia:Teahouse/Questions]] [21:47:03] ahh [21:47:09] So if we sum, we're counting links, not postings. [21:47:28] how about a chart with just links to Wikipedia:Teahouse then? [21:47:32] Seems to me that we should count all postings that include links to [[WP:Teahouse]] and [[Wikipedia:Teahouse]] [21:47:33] Yeah [21:48:03] so, in your current dataset, they're not associated with a rev_id? [21:48:08] "they" = each link [21:48:17] They are, but one rev_id can have many links. [21:48:30] We can just count distinct rev_ids. [21:48:32] can't we just count distinct rev_ids that contain at least 1 link? [21:48:41] hehe. yes! [21:49:03] This'll pick up all posting on User_talk regardless of whether it is an invite for a newcomer/host or just discussion. [21:49:22] that's okay with me. [21:49:35] the overall point is that the Teahouse has become an integral part of the wiki-ecology. [21:50:20] kk [21:50:29] * halfak does pointy statistics. [21:52:27] Hmm... Looks like there were a lot of links to [[WP:TH]] for a while [21:52:32] In 2007 [21:53:06] Hmm... Weird. I don't see a deletion history there. [21:53:14] So maybe it is a common typo. [21:53:19] "a lot" = 153 [21:53:42] Which compares to ~13k per month now [21:53:52] * halfak sanity checks [21:55:00] gotta be a typo [21:59:52] Ugh. Looks like I might have a problem in my code. [22:00:00] Check this out: https://en.wikipedia.org/w/index.php?diff=64816177 [22:00:04] * halfak fixes regexes. [22:02:12] halfak, we're research meeting! [22:52:35] leila: are you in the office for the open drinks thing? [22:52:53] unfortunately, not, YuviPanda. :-( [22:52:59] :( [22:53:07] madhuvishy: are you? [22:53:25] leila: btw, have I shown you tmpnb.org? [22:53:48] nooo [22:53:50] lemme check [22:54:15] doh! what is this YuviPanda? :D [22:55:46] leila: it's basically a way to allow people to do computation / research without them having to deal with ssh, commandline, etc :) [22:56:09] sign up / sign in, and bam - python, R, julia notebooks, or just a terminal (try the 'new' button) you can hit from the web [22:56:14] that's cool, YuviPanda. [22:56:45] I'm setting one up on toollabs that will be open to users, similar to Quarry. [22:56:47] * guillom has just started reading up on iPython, pandas, etc. [22:57:00] wanted people to do some actual real research-type stuff with it too at some point to identify pain points [22:57:02] nice guillom [22:57:11] YuviPanda: Setting up a jupyter instance in labs? [22:57:28] guillom: yup. with OAuth :) [22:57:32] \o/ [22:58:00] I know at least Stephen LaPorte and Ellery are familiar with this. Not sure who else has been actively using this tool. [22:58:04] guillom: specifically, I'm doing this in toollabs as a test (and demo) of the new clustering tool we're setting up (kubernetes) [22:58:10] cool [22:58:43] guillom: I shall keep you posted if you don't mind :D I need a few test subjects to whitelist first before unleashing the gates, etc :) [22:59:15] YuviPanda: I don't mind at all! Although, as I mentioned, I'm just starting out so I'm not a great typical user. [22:59:34] Presumably halfa.k is also familiar with it. [22:59:52] guillom: sure, but I think the power of this is making it easy for people to start up and go forward :D steven and ellery have used it before a lot, so I guess maybe less typical than the people I'm trying to target maybe :D [23:00:05] heh, fair enough :) [23:00:18] \o/ cool [23:02:50] YuviPanda: what open drinks [23:03:07] madhuvishy: dartar announced a few weeks ago in wmfsf [23:03:20] madhuvishy: http://www.meetup.com/SFOpenDrinks/events/226348327/ [23:03:31] madhuvishy: http://www.meetup.com/SFOpenDrinks/events/226348327/ [23:03:31] Today at 6pm [23:03:33] hah :) [23:03:35] :p [23:04:07] * guillom afk again. [23:05:13] YuviPanda: ya i'll be here for a while, might leave at 7 ish though [23:05:17] let me rsvp [23:05:22] ok [23:05:42] the eventbrite might've closed but IIRC that's ok for you since you already have a keycard [23:23:11] Wish I could drink openly in SF. Have Fun! [23:33:11] hey halfak, do you happen to know if this experiment was ever run? https://meta.wikimedia.org/wiki/Research:Task_recommendations/Experiment_one [23:33:57] J-Mo, yes, it was. [23:34:09] The Growth Team imploded before I wrote up the results. [23:34:23] womp-womp [23:34:25] See https://meta.wikimedia.org/wiki/Research_talk:Task_recommendations#Work_log [23:34:36] TL;DR: no effect that I could measure. [23:36:43] yeah [23:37:00] but at least you got some angry metapedians to badger you, halfak. Yay? [23:37:25] "I think it is not WMFs job to give tasks to users. It is WMFs job to give tools to users to choose their own tasks. " [23:37:42] anyway… back to work. Thanks halfak! [23:59:07] :) [23:59:43] * halfak values the metapedians thoughts, but may choose to not engage.