[10:59:26] (03Abandoned) 10Ladsgroup: Move all of ORES preferences to a dedicated tab [extensions/ORES] - 10https://gerrit.wikimedia.org/r/361298 (https://phabricator.wikimedia.org/T167910) (owner: 10Ladsgroup) [15:54:47] tgr|away: I see that one user has claimed both of the microtasks you wrote…. Should we ask developers to not self-assign those tasks, to make space for multiple approaches? [18:41:23] halfak: got some minutes? [18:42:32] at a workshop. But I'll try to read an respond quickly, [18:43:33] oh, its regarding the nature of WikiProjects parsing, but you can choose to reply later if needed [18:44:12] basically iterative vs recursive... [18:44:43] what i have at my disposal is sections list like - https://en.wikipedia.org/w/api.php?action=parse&page=Wikipedia:WikiProject_Council/Directory/Science&prop=sections [18:45:45] any parser iterative or resursive goes over this array, fetching each section text using api and inserting it into the data structure [18:45:55] *would go [18:47:22] anyway, i'll take this up in next sync up [18:49:23] codezee, I'm OK with the recursive strategy for now. Let's get it merged and if we find time later, we can revisit it. Most importantly, this works. :) [18:50:18] yeah, that is true, when this gets in i'll write a small PR to handle request failures gracefully then we'll be good to close that task :) [18:57:42] If you write the recursion as a sequence generator, it can be interfaced with as part of a linear pipeline. [18:58:49] codezee: cool! I’ll try to code review whenever you write those. [18:59:34] awight: any example of a recursion problem as a seq generator that i can look at? [19:11:45] awight: just out of curiosity, look at https://en.wikipedia.org/wiki/Wikipedia:WikiProject_Council/Directory/Culture#Music there, the entry links to an exactly similar format new page whose info has to be inserted in its parent structure [19:12:15] not knowing the number of such pages we need to parse beforehand, makes this challenging [19:12:48] codezee: Harrgh, that’s a special type of recursion, with table nesting inside of document nesting! Scary stuff you’re unearthing. [19:13:18] it reminds me of the mediawiki parser itself which is way more complex than this and still works :P [19:13:27] o/ [19:15:35] Zppix: o/ [19:20:16] awight: I think self-assignment is fine, but only one task of course [19:20:53] multiple people doing the same microtask seems awkward because the second one can just check what the first one did and have an unfair advantage [19:21:20] maybe devs can choose to work together, though? [19:21:43] tgr: i disagree devs working together on any task can reduce number of errors [19:23:00] well, for one thing only one of them will get selected (if any - apparently we have 6 tasks but only 4 spots) which is not a great setup for collaboration [19:23:38] and in any case, how would you select the best candidate based on a collaboritve patch? [19:23:45] tgr: above discussion for any competition? [19:24:05] Outreachy [19:24:18] oh, which task is being refered to? [19:25:43] oh i see now [19:29:48] the captcha one is an interesting project! [19:31:52] if i get it right its basically trying to come up with useful latent metrics associated with a user's behaviour like clicks to distinguish bots from humans, right? [19:32:06] to begin with atleast [19:53:06] codezee: yeah, that's the idea [21:05:45] awight: did that JADE stakeholders list come together, that you know off? [21:05:51] *of [21:06:43] Keegan: hi! I haven’t dug into the community contributors, so it’s probably unchanged, if you’re already seen it? https://www.mediawiki.org/wiki/ORES/Stakeholders [21:06:49] *you’ve [21:08:42] Okay, I'm looking for the wiki liaisons then. I can put something together myself, no problem. I was under the impression that such a collection might already exist. [21:09:15] If not, no prob. Have it together by end of day Monday to contact people on Tuesday [21:09:19] I would love to help, I just got distracted by the schema and use case stuff [21:09:22] {{File:Jade_data_flow_diagram.svg}} [21:09:22] 10[2] 04https://meta.wikimedia.org/wiki/Template:File:Jade_data_flow_diagram.svg [21:09:29] not that one :) [21:09:31] [[File:Jade_data_flow_diagram.svg]] [21:09:32] 10[3] 04https://meta.wikimedia.org/wiki/File:Jade_data_flow_diagram.svg [21:09:41] No problem at all, I can handle it ;) [21:10:24] I'll put together some text for y'all to review on Monday while I put the list together. [21:10:48] https://phabricator.wikimedia.org/project/board/1306/query/mp_A9.2LdMjO/ [21:11:28] Here’s a more complete list, https://phabricator.wikimedia.org/project/board/1901/query/mp_A9.2LdMjO/ [21:11:38] Very helpful, thank you [21:14:02] Paste me the doc URL if you want to collaborate, I have a bit of time today [21:14:49] awight: let me know if you need me to merge any prs [21:15:19] awight: I haven't written anything yet. One more meeting then my day is winding down for the weekend. [21:15:37] I'll link on Monday morning. Won't take long. [21:15:54] Zppix: This isn’t ready to merge, but feel free to review the PDF in https://github.com/wiki-ai/ores-diagrams/pull/2 [21:16:09] +1 for having a weekend! [21:17:07] looking now [21:18:12] awight: i like it but the start point is a little confusing to find [21:18:33] yeah huh—there are two start points [21:18:54] the Reviewer sees stuff to do, and the Researcher curates sets of artifacts needing review. [21:19:42] awight: maybe that should be more clear as i didnt even see that there was 2 it took me a few mins just to find 1 [21:20:55] Data flow can be weird that way though—there are external interfaces, then in my experience lots of feedback cycles within each system. [21:21:03] Sometimes it really is a chicken and an egg.