[07:02:58] 10Jade, 10Scoring-platform-team: Add text to error when submitting a label that already exists. - https://phabricator.wikimedia.org/T247450 (10kevinbazira) a:03kevinbazira [07:03:34] 10Jade, 10Scoring-platform-team (Current): Add text to error when submitting a label that already exists. - https://phabricator.wikimedia.org/T247450 (10kevinbazira) [07:23:41] kevinbazira: we talked about a TDD and a refactoring workshops. Next sessions are scheduled: https://www.mediawiki.org/wiki/Code_Health_Group/projects/DevEd/Workshops [07:23:57] cc: halAFK ^ [07:24:12] Thanks gehel! [14:27:59] 10Scoring-platform-team, 10Research: Write Python util for converting Wikidata claims to features for ML models - https://phabricator.wikimedia.org/T252775 (10Isaac) [14:28:58] halfak: ^^ i was a bit slow on creating this phab task but if you get a chance before our Outreachy meeting today, maybe take a look to make sure I captured your thinking about this first step [14:29:36] i also am creating an agenda for our meeting today: https://etherpad.wikimedia.org/p/DibyaOutreachy20 [14:29:46] What does "checking references" mean? [14:50:23] isaacj, ^ [14:51:20] frankly i don't think there are any statement-level filters at this moment, but i could imagine for instance only wanting the statement with highest rank or only wanting statements with references associated (hence "checking references" though that's very broad language) [14:51:26] halfak: ^^ [14:51:57] https://en.wikipedia.org/wiki/You_aren%27t_gonna_need_it [14:52:11] Let's build the feature when we're ready to use it :) [14:53:27] haha :thumbs up: [15:16:36] 10Scoring-platform-team (Current), 10Technical-blog-posts: Story idea for blog: Building algorithmic systems while keeping community in the loop - https://phabricator.wikimedia.org/T251426 (10srodlund) @Halfak Can you confirm the co-authors for this, so I can make sure they are credited on the post? I will pl... [16:01:53] 10Scoring-platform-team (Current), 10Technical-blog-posts: Story idea for blog: Building algorithmic systems while keeping community in the loop - https://phabricator.wikimedia.org/T251426 (10Halfak) Author line is "C. Estelle Smith". Here's Estelle's website for linking: https://colleenestellesmith.com/ Oth... [16:09:57] 10Scoring-platform-team (Current), 10Discovery-Search, 10Elasticsearch, 10revscoring, 10artificial-intelligence: Improve the performance and quality of tokenization in revscoring - https://phabricator.wikimedia.org/T248480 (10HAKSOAT) I checked the `regex` package. While it works great for the unicode sc... [17:32:28] 10Scoring-platform-team (Current), 10Technical-blog-posts: Story idea for blog: Building algorithmic systems while keeping community in the loop - https://phabricator.wikimedia.org/T251426 (10srodlund) @Halfak this is published: https://twitter.com/wikimediatech/status/1260986011317760000 :-) [17:43:56] It's alive! https://techblog.wikimedia.org/2020/05/14/what-does-it-mean-to-keep-community-in-the-loop-when-building-algorithms-for-wikipedia/ [17:44:20] Nice note to go to lunch on. [20:24:04] posting our async update -- [20:24:18] kevinbazira- [20:24:23] Y: [20:24:25] Removed mousedown event handling from Jade ListWidgets [20:24:27] The FacetListWidget, ProposalListWidget and EndorsementListWidget all inherit from the OOUI SelectWidget Class. [20:24:29] The OOUI SelectWidget hijacks mouse events as seen here: [20:24:31] https://github.com/wikimedia/oojs-ui/blob/413a86cf206aa2ec716a203e2381f338c96a0c5e/src/widgets/SelectWidget.js#L73-L78 [20:24:33] This prevents users from using the mouse to set cursor position in text inputs and textareas when editing endorsement comments and proposal notes. [20:24:35] Since the aforementioned Jade elements that inherit from the SelectWidget don't rely on the mousedown event handler. I have removed the mousedown event handler from them. [20:24:37] Now text inputs and textareas within these Jade elements can receive the mousedown event and users can set the cursor position in text inputs and textareas === Better UX. [20:24:39] T: [20:24:41] Fixed label proposal error message [20:24:43] Started looking into Jade 2ndary UI integrations [20:24:45] halfak- [20:24:47] Y: I set up the hiwiki labeling campaign. I discussed and added information to nlwiki articlequality modeling. I met with a coalition looking at citation modeling -- no clear directions there. I discussed chtnnh's work on generating weighted sum measurements for model auditing. https://gist.github.com/chtnnh/66e150006dc12f45ade28f303b68c9a6 I also finally deployed wikilabels UI improvements. [20:24:49] T: Met with Dibya -- a student working on topic modeling for Wikidata. I reviewed a small change by kevinbazira to an i18n message. I'm going to write a case study about our work with ptwiki for the ORES paper. [20:25:06] haksoat- [20:25:09] Y: Meeting with Aaron to talk about tokenization progress. Profiling of regex library which turns out to be slower. Tried looking at the codebase to see if we can pick up the part for unicode scripts alone, but I can see a lot of C code which makes me wonder why it is slower. Looking for a way to get the unicode scripts without the library. [20:25:11] T: Working on using a Trie to reimplement the regex for English idioms. I was able to do first level factorization earlier in the day and I got some form of improvement. Really excited to see how this works out. [20:25:22] and me- [20:25:24] Y: Reviewed Kevin's ListWidget mousedown patchset. Continued with the ORES blubberfile and can now generate valid multi-stage Dockerfiles used to spin up a basic ORES container. Also set up a small kubeflow instance using microk8s and started attempting to wrap a revscoring model using seldon-core. [20:25:26] T: Do more testing with ORES blubber file and try to use it on the ORES deploy repo, also will continue working on deploying a revscoring model onto microk8s and looking at the kf pipeline sdk