[09:38:32] 06Revision-Scoring-As-A-Service, 13Patch-For-Review: What's going on with ORES logs? - https://phabricator.wikimedia.org/T148436#2747821 (10akosiaris) 05Open>03Resolved a:03akosiaris So, a 6month old patch in https://gerrit.wikimedia.org/r/285615 by me fixed the logrotate bug. That bug had not yet caused... [11:07:10] 06Revision-Scoring-As-A-Service, 13Patch-For-Review: What's going on with ORES logs? - https://phabricator.wikimedia.org/T148436#2722963 (10Ghassanmas) The log pattern specifically "how the long of each line varies as you scroll down" looks like the historical map of Palestine but turned exactly 180 degree on... [12:32:49] 06Revision-Scoring-As-A-Service, 10MediaWiki-extensions-ORES, 15User-Ladsgroup: Visually report damaging confidence - https://phabricator.wikimedia.org/T144922#2614593 (10Pginer-WMF) In the recent iteration on the exploration we are doing to integrate ORES filters and others into Recent Changes (T147632), a... [14:28:35] 10Revision-Scoring-As-A-Service-Backlog, 10ORES: Meta ORES: UI for reviewing how ORES classifies you and your stuff - https://phabricator.wikimedia.org/T148700#2748356 (10Halfak) p:05Triage>03Low [14:29:26] 10Revision-Scoring-As-A-Service-Backlog, 10ORES: Implement selective purging of model scores in varnish - https://phabricator.wikimedia.org/T148999#2748358 (10Halfak) p:05Triage>03Low [14:31:37] 06Revision-Scoring-As-A-Service, 10ORES: Clean up ORES service documentation - https://phabricator.wikimedia.org/T148974#2748361 (10Halfak) [14:31:39] 10Revision-Scoring-As-A-Service-Backlog, 10ORES, 10revscoring: Build a sketch of basic ORES structure - https://phabricator.wikimedia.org/T148692#2748360 (10Halfak) [14:32:12] 10Revision-Scoring-As-A-Service-Backlog, 10ORES, 10revscoring: Build a sketch of basic ORES model building patterns - https://phabricator.wikimedia.org/T148692#2730166 (10Halfak) [14:32:22] 10Revision-Scoring-As-A-Service-Backlog, 10ORES, 10revscoring: Build a sketch of basic ORES model building patterns - https://phabricator.wikimedia.org/T148692#2730166 (10Halfak) p:05Triage>03Low [14:37:26] 10Revision-Scoring-As-A-Service-Backlog, 10ORES: Implement parallel connection limit for querying ORES - https://phabricator.wikimedia.org/T148997#2748373 (10Halfak) [14:37:46] 06Revision-Scoring-As-A-Service, 10ORES: Clean up ORES service documentation - https://phabricator.wikimedia.org/T148974#2748374 (10Halfak) a:03Halfak [14:38:30] 10Revision-Scoring-As-A-Service-Backlog, 10AbuseFilter, 10MediaWiki-extensions-ORES, 15User-Ladsgroup: [Spike] Investigate building a hook for abuse filter - https://phabricator.wikimedia.org/T123178#2748375 (10Halfak) [14:39:26] 06Revision-Scoring-As-A-Service, 10ChangeProp, 10EventBus, 06Services (watching), 15User-mobrovac: Rewrite ORES precaching change propagation configuration as a code module - https://phabricator.wikimedia.org/T148714#2748381 (10Halfak) @awight, what do you think of ^? [14:40:01] 10Revision-Scoring-As-A-Service-Backlog, 10ORES: Implement prioritization of request processing - https://phabricator.wikimedia.org/T148594#2748383 (10Halfak) a:05Halfak>03None [14:42:30] 10Revision-Scoring-As-A-Service-Backlog, 10ORES: Implement parallel connection limit for querying ORES - https://phabricator.wikimedia.org/T148997#2748402 (10Halfak) p:05Triage>03Normal [14:42:53] 10Revision-Scoring-As-A-Service-Backlog, 10ORES: Implement prioritization of request processing - https://phabricator.wikimedia.org/T148594#2748405 (10Halfak) p:05Triage>03Normal [16:49:39] o/ halfak [16:49:55] o/ [16:50:04] In meetings will be back in 40 mins [16:50:17] kk [17:28:50] halfak: I'm back. One thing. ORES instances in labs will be rebooted because of the "dirty cow" security hole [17:29:47] wat [17:32:30] Out of meeting! [17:32:33] sabya, [17:32:47] So yeah, thinking about next steps with semi-supervised learning. [17:32:47] o/ [17:32:54] 1. Read up on methods. [17:32:56] halfak: https://lists.wikimedia.org/pipermail/labs-announce/2016-October/000173.html [17:33:14] 2. Take our labeled data for damaging/not and split into train/test set [17:33:21] 3. Build model on training set. [17:33:45] https://www.theguardian.com/technology/2016/oct/21/dirty-cow-linux-vulnerability-found-after-nine-years [17:33:51] 4. Run model against a random sample of revisions and take the revisions that are strongly scored (high confidence of "damaging"/not) [17:34:08] 5. Train a new model on the training set + the strongly-labeled observations. [17:34:18] 6. Test against the test set and see if we do better. [17:34:45] cow = copy-on-write lol [17:35:17] ok, regarding #1, what methods should I read up? [17:35:47] https://en.wikipedia.org/wiki/Semi-supervised_learning [17:36:46] sabya, we could probably get you a copy of https://www.amazon.com/Semi-Supervised-Learning-Adaptive-Computation-Machine/dp/0262514125 [17:39:52] got it [17:45:16] Nice! [17:45:50] I'm really curious what you learn from this. It might be a good route to writing a journal paper. [17:46:04] I know Amir1 is interested in semi-supervised learning too. [17:46:53] Yeah, I'm super excited I wish I had more time [17:46:55] :( [17:50:34] this is gonna be exciting. will order the book. [17:55:22] halfak: ordering the book. is that the best book I could buy on this subject? [17:57:22] sabya, really not sure [17:57:23] :( [17:57:54] sabya, I think we might be able to reimburse you. Why don't you read what's freely available for now while I check. [17:58:41] sure. sounds great. [18:03:37] bed time. o/ [18:40:05] halfak: https://github.com/wiki-ai/grafana-dashboards [18:40:51] Nice [18:48:01] halfak: It is possible to update grafana dashboards with API: I need to write the update.php script: http://docs.grafana.org/http_api/dashboard/ [18:51:31] 06Revision-Scoring-As-A-Service, 15User-Ladsgroup: Revscoring grafana dashboards in a repo - https://phabricator.wikimedia.org/T149347#2749461 (10Ladsgroup) [18:52:35] 06Revision-Scoring-As-A-Service, 15User-Ladsgroup: Write script to update grafana dashboards using API - https://phabricator.wikimedia.org/T149348#2749476 (10Ladsgroup) [18:54:55] 06Revision-Scoring-As-A-Service, 15User-Ladsgroup: Write script to update grafana dashboards using API - https://phabricator.wikimedia.org/T149348#2749506 (10Ladsgroup) Notes on how to implement that: http://docs.grafana.org/http_api/dashboard/ [19:38:46] 06Revision-Scoring-As-A-Service, 10rsaas-articlequality : Explore quality prediction valleys for en:Biology - https://phabricator.wikimedia.org/T149354#2749662 (10Halfak) [19:39:53] 06Revision-Scoring-As-A-Service, 10rsaas-articlequality : Explore quality prediction valleys for en:Biology - https://phabricator.wikimedia.org/T149354#2749675 (10Halfak) https://meta.wikimedia.org/wiki/Research_talk:Automated_classification_of_article_quality/Work_log/2016-10-27 My summary: large removals (m... [19:50:19] halfak: around? I have a quick question [19:50:34] 10Revision-Scoring-As-A-Service-Backlog, 07Spike: [Spike] Semi-supervised machine learning - https://phabricator.wikimedia.org/T143123#2749758 (10Halfak) I talked to @Sabya in IRC. Here's the steps that I recommended. 1. Read up on methods. 2. Take our labeled data for damaging/not and split into train/test... [20:35:29] I ask later [20:35:30] o/ [21:28:41] Amir1, sorry running fast for blog stuff