Page MenuHomePhabricator

ellery (ewulczyn)
Disabled

Projects

User Details

User Since
Dec 9 2014, 6:35 PM (485 w, 2 d)
Roles
Disabled
LDAP User
Ewulczyn
MediaWiki User
Unknown
This account has been disabled.

Recent Activity

Feb 25 2018

Chicocvenancio awarded T144234: Detox notebooks on PAWS a Love token.
Feb 25 2018, 9:33 PM · PAWS, Discussion-modeling (of Toxicity)

Feb 21 2017

ellery added a comment to T158472: Search by categories in GapFinder - Translate.

@leila We could talk to Magnus about how he does category search in https://tools.wmflabs.org/not-in-the-other-language. Also, this may be very easy to implement by using WDQS now ...

Feb 21 2017, 7:25 PM · Recommendation-API, GapFinder

Feb 14 2017

ellery added a comment to T157724: Cluster Access for Nithum Thain.

That would be @DarTar .

Feb 14 2017, 5:48 PM · Patch-For-Review, Research, SRE, SRE-Access-Requests

Feb 9 2017

ellery added a comment to T157724: Cluster Access for Nithum Thain.

Thanks @RobH. Nithum signed an NDA that was approved by Manprit, Dario and Wes. I pointed Nithum to this ticket and asked him to complete the tasks you listed. The access group should be analytics-privatedata-users.

Feb 9 2017, 10:54 PM · Patch-For-Review, Research, SRE, SRE-Access-Requests
ellery updated the task description for T157724: Cluster Access for Nithum Thain.
Feb 9 2017, 10:49 PM · Patch-For-Review, Research, SRE, SRE-Access-Requests
ellery added a project to T157724: Cluster Access for Nithum Thain: Research.
Feb 9 2017, 8:00 PM · Patch-For-Review, Research, SRE, SRE-Access-Requests
ellery created T157724: Cluster Access for Nithum Thain.
Feb 9 2017, 8:00 PM · Patch-For-Review, Research, SRE, SRE-Access-Requests

Feb 7 2017

Abbe98 awarded T144234: Detox notebooks on PAWS a Like token.
Feb 7 2017, 7:31 PM · PAWS, Discussion-modeling (of Toxicity)

Feb 6 2017

ellery created T157371: Add option to rank by number of languages the article exists in.
Feb 6 2017, 7:21 PM · Recommendation-API

Jan 30 2017

ellery added a comment to T156522: Logging for AB testing recommender systems for CX Suggestions.

Yes

Jan 30 2017, 5:59 PM · MW-1.29-release (WMF-deploy-2017-02-28_(1.29.0-wmf.14)), Language-2017 Sprint 3, Language-2017 Sprint 2, Language-team January-March 2017, ContentTranslation, GapFinder

Jan 27 2017

ellery added a comment to T153443: Create user interface for related articles recommendations.

I set up a prototype on tool labs that consumes the related articles API.

Jan 27 2017, 11:49 PM · GapFinder
ellery updated the task description for T156522: Logging for AB testing recommender systems for CX Suggestions.
Jan 27 2017, 8:08 PM · MW-1.29-release (WMF-deploy-2017-02-28_(1.29.0-wmf.14)), Language-2017 Sprint 3, Language-2017 Sprint 2, Language-team January-March 2017, ContentTranslation, GapFinder
ellery created T156522: Logging for AB testing recommender systems for CX Suggestions.
Jan 27 2017, 8:05 PM · MW-1.29-release (WMF-deploy-2017-02-28_(1.29.0-wmf.14)), Language-2017 Sprint 3, Language-2017 Sprint 2, Language-team January-March 2017, ContentTranslation, GapFinder

Jan 23 2017

ellery added a comment to T148843: Remove computational bottlenecks in stats machine via adding a GPU that can be used to train ML models.

I'm in no rush, especially if I can get some budget to rent GPUs on AWS in the meantime.

Jan 23 2017, 11:56 PM · Analytics-Radar, Patch-For-Review, User-Elukey, SRE, Research-management

Jan 18 2017

ellery created T155591: Build/Port Frontend for semantically related article API .
Jan 18 2017, 12:41 AM · GapFinder

Dec 15 2016

ellery moved T139704: Discussion modeling - release notebooks; write up and present results from In progress to Done on the Discussion-modeling (of Toxicity) board.
Dec 15 2016, 11:10 PM · Epic, Research-and-Data-2016-17-Q1, Discussion-modeling (of Toxicity), Research-Freezer
ellery moved T143707: WWW Paper from In progress to Done on the Discussion-modeling (of Toxicity) board.
Dec 15 2016, 11:10 PM · Discussion-modeling (of Toxicity)
ellery created T153379: Newcomer Good Faith Model .
Dec 15 2016, 11:10 PM · Discussion-modeling (of Toxicity)
ellery moved T144065: Put EDP models on Demo from Backlog to Done on the Discussion-modeling (of Toxicity) board.
Dec 15 2016, 11:08 PM · Discussion-modeling (of Toxicity)
ellery closed T144065: Put EDP models on Demo as Declined.
Dec 15 2016, 11:08 PM · Discussion-modeling (of Toxicity)
ellery moved T143710: Detox Dashboard from Backlog to Done on the Discussion-modeling (of Toxicity) board.
Dec 15 2016, 11:08 PM · Epic, Discussion-modeling (of Toxicity)
ellery closed T143710: Detox Dashboard as Declined.
Dec 15 2016, 11:07 PM · Epic, Discussion-modeling (of Toxicity)
ellery moved T127533: Detox Data Release from Backlog to Staged on the Discussion-modeling (of Toxicity) board.
Dec 15 2016, 11:07 PM · Epic, Documentation, Discussion-modeling (of Toxicity)
ellery closed T143701: Switch from keras to tflearn for deeplearning as Invalid.
Dec 15 2016, 11:07 PM · Discussion-modeling (of Toxicity)
ellery moved T143701: Switch from keras to tflearn for deeplearning from Backlog to Done on the Discussion-modeling (of Toxicity) board.
Dec 15 2016, 11:07 PM · Discussion-modeling (of Toxicity)
ellery moved T146186: Detox and Bias from Backlog to Staged on the Discussion-modeling (of Toxicity) board.
Dec 15 2016, 11:06 PM · Epic, Discussion-modeling (of Toxicity)
ellery moved T151502: C-level detox presentation from Backlog to Done on the Discussion-modeling (of Toxicity) board.
Dec 15 2016, 11:06 PM · Research-Archive, Discussion-modeling (of Toxicity)
ellery moved T127531: Harassment and User Retention from Backlog to Staged on the Discussion-modeling (of Toxicity) board.
Dec 15 2016, 11:06 PM · Epic, Discussion-modeling (of Toxicity)
ellery moved T149132: Add annotator data to figshare from Staged to Done on the Discussion-modeling (of Toxicity) board.
Dec 15 2016, 11:05 PM · Discussion-modeling (of Toxicity)
ellery moved T149133: Add models to figshare from Staged to Done on the Discussion-modeling (of Toxicity) board.
Dec 15 2016, 11:05 PM · Discussion-modeling (of Toxicity)
ellery moved T149135: Make detox python package for model deserialization from Staged to Done on the Discussion-modeling (of Toxicity) board.
Dec 15 2016, 11:05 PM · Discussion-modeling (of Toxicity)
ellery moved T149134: Add comment corpus to figshare from Staged to Done on the Discussion-modeling (of Toxicity) board.
Dec 15 2016, 11:05 PM · Discussion-modeling (of Toxicity)
ellery moved T149131: Put paper on arxiv from Staged to Done on the Discussion-modeling (of Toxicity) board.
Dec 15 2016, 11:05 PM · Discussion-modeling (of Toxicity)

Dec 9 2016

ellery assigned T152750: Satisfying requests for n recs to schana.
Dec 9 2016, 12:16 AM · GapFinder
ellery created T152750: Satisfying requests for n recs.
Dec 9 2016, 12:16 AM · GapFinder

Dec 8 2016

ellery created T152745: Remove pageview counts from API calls .
Dec 8 2016, 11:56 PM · GapFinder
ellery moved T144285: Measure Impact of Recommendations on CX from Next Up to Done on the GapFinder board.
Dec 8 2016, 11:54 PM · GapFinder
ellery moved T139785: Log whether a CX translation was initiated via GapFinder from Next Up to Done on the GapFinder board.
Dec 8 2016, 11:54 PM · GapFinder

Dec 2 2016

ellery added a comment to T148843: Remove computational bottlenecks in stats machine via adding a GPU that can be used to train ML models.

@RobH Thank you for the thorough investigation :). Now we know that the stat machines cannot accommodate a top-of-the-line GPU. That being said, there are many different options. Looking at what Nvidia has on offer, do you have a sense of what the most powerful model we can accommodate is?

Dec 2 2016, 6:03 PM · Analytics-Radar, Patch-For-Review, User-Elukey, SRE, Research-management

Nov 30 2016

ellery updated the task description for T148453: Research Offsite 16-17 planning.
Nov 30 2016, 5:45 PM · Design-Research, Research-management, Research

Nov 14 2016

ellery added a comment to T148843: Remove computational bottlenecks in stats machine via adding a GPU that can be used to train ML models.

@elukey, DC-Ops Do you have an estimated timeline for this task?

Nov 14 2016, 7:16 PM · Analytics-Radar, Patch-For-Review, User-Elukey, SRE, Research-management

Nov 11 2016

ellery updated the task description for T150488: Research Staff "VC" special edition.
Nov 11 2016, 6:47 PM · Research-Archive, Research-management

Oct 31 2016

ellery added a comment to T148843: Remove computational bottlenecks in stats machine via adding a GPU that can be used to train ML models.

This is the GPU we would like to order.

Oct 31 2016, 5:24 PM · Analytics-Radar, Patch-For-Review, User-Elukey, SRE, Research-management

Oct 26 2016

ellery added a comment to T143694: Preliminary Design document for A/B testing.

The pseudo code does not quite match the current text description of the Double Bucket proposal.

Oct 26 2016, 8:44 PM · Reading-Admin, Analytics-Kanban, Performance-Team, Traffic, SRE
ellery moved T139785: Log whether a CX translation was initiated via GapFinder from Done to Next Up on the GapFinder board.
Oct 26 2016, 8:34 PM · GapFinder
ellery added a comment to T139785: Log whether a CX translation was initiated via GapFinder .

Is there a way to link events from log.ContentTranslationCTA_11616099 to wikishared.cx_translations. At a high level, I want to see which individual translations where started from our tool.

Oct 26 2016, 8:33 PM · GapFinder
ellery reopened T139785: Log whether a CX translation was initiated via GapFinder as "Open".
Oct 26 2016, 8:25 PM · GapFinder
ellery added a comment to T148843: Remove computational bottlenecks in stats machine via adding a GPU that can be used to train ML models.

Yes, I was operating under the assumption that stat1004 was the local "compute" node and that stat1002 is more or less reserved for Zachte.

Oct 26 2016, 6:37 PM · Analytics-Radar, Patch-For-Review, User-Elukey, SRE, Research-management
ellery added a comment to T148843: Remove computational bottlenecks in stats machine via adding a GPU that can be used to train ML models.

@elukey I have a slight preference for stat1004 since it has access to HDFS

Oct 26 2016, 5:45 PM · Analytics-Radar, Patch-For-Review, User-Elukey, SRE, Research-management

Oct 25 2016

ellery created T149136: Update wikidetox appspot model.
Oct 25 2016, 9:53 PM · Discussion-modeling (of Toxicity)
ellery created T149135: Make detox python package for model deserialization.
Oct 25 2016, 9:50 PM · Discussion-modeling (of Toxicity)
ellery created T149134: Add comment corpus to figshare.
Oct 25 2016, 9:49 PM · Discussion-modeling (of Toxicity)
ellery created T149133: Add models to figshare.
Oct 25 2016, 9:49 PM · Discussion-modeling (of Toxicity)
ellery created T149132: Add annotator data to figshare.
Oct 25 2016, 9:49 PM · Discussion-modeling (of Toxicity)
ellery created T149131: Put paper on arxiv.
Oct 25 2016, 9:48 PM · Discussion-modeling (of Toxicity)

Oct 21 2016

ellery updated the task description for T148453: Research Offsite 16-17 planning.
Oct 21 2016, 3:35 AM · Design-Research, Research-management, Research

Oct 17 2016

ellery updated the task description for T145001: Research showcase October 2016.
Oct 17 2016, 9:15 PM · Research-Archive, Research-outreach
ellery updated the task description for T145001: Research showcase October 2016.
Oct 17 2016, 9:14 PM · Research-Archive, Research-outreach
ellery updated the task description for T145001: Research showcase October 2016.
Oct 17 2016, 9:14 PM · Research-Archive, Research-outreach

Oct 13 2016

ellery added a comment to T139704: Discussion modeling - release notebooks; write up and present results.

First draft of the paper is complete.

Oct 13 2016, 4:17 PM · Epic, Research-and-Data-2016-17-Q1, Discussion-modeling (of Toxicity), Research-Freezer

Oct 12 2016

ellery added a comment to T147708: Facilitate Wikidev'17 main topic "Artificial Intelligence to build and navigate content".

@Halfak I'm happy to help out. What would be the best way for me to contribute. I'm happy to help out with facilitating discussions, doing some demo's, or giving a technical talk ...

Oct 12 2016, 1:54 AM · Wikimedia-Developer-Summit

Oct 11 2016

ellery added a comment to T145001: Research showcase October 2016.

sent email on logistics to speakers yesterday

Oct 11 2016, 8:24 PM · Research-Archive, Research-outreach

Oct 7 2016

ellery updated the task description for T144284: [Placeholder] GapFinder Continued Engagement.
Oct 7 2016, 10:59 PM · Epic, GapFinder
ellery updated the task description for T144284: [Placeholder] GapFinder Continued Engagement.
Oct 7 2016, 10:58 PM · Epic, GapFinder

Sep 30 2016

ellery updated the task description for T144576: Make sure each team member with published research outputs has an ORCID.
Sep 30 2016, 11:53 PM · Research-Archive, Research-management
ellery updated the task description for T144576: Make sure each team member with published research outputs has an ORCID.
Sep 30 2016, 11:53 PM · Research-Archive, Research-management

Sep 20 2016

ellery removed a project from T143707: WWW Paper: Epic.
Sep 20 2016, 6:51 PM · Discussion-modeling (of Toxicity)
ellery moved T143707: WWW Paper from Staged to In progress on the Discussion-modeling (of Toxicity) board.
Sep 20 2016, 6:51 PM · Discussion-modeling (of Toxicity)
ellery added a comment to T127531: Harassment and User Retention.

We have models for personal attacks and aggression already. We have data on civility.

Sep 20 2016, 6:50 PM · Epic, Discussion-modeling (of Toxicity)
ellery created T146186: Detox and Bias.
Sep 20 2016, 6:49 PM · Epic, Discussion-modeling (of Toxicity)
ellery created T146185: Detox Visualization.
Sep 20 2016, 6:47 PM · Epic, Discussion-modeling (of Toxicity)
ellery added a comment to T135762: A/B Testing solid framework .

@Nuria I certainly don't disagree that segmentation must be done at the user level. I'm saying that the test statistics (or metrics as you are calling them) also need to be computed at a user level (i.e. compare the average number of clicks per user between treatment and control instead of just comparing the number of clicks across all users between treatment and control). To do this, there needs to be some way of grouping data by user in each experiment. The current proposal is missing a mechanism to achieve this.

Sep 20 2016, 7:16 AM · Analytics-Radar, SRE, Traffic
ellery added a comment to T135762: A/B Testing solid framework .

@Neil_P._Quinn_WMF I'm saying that for any online AB test you to be able to group the experimental data by user. The proposed framework does not provide a mechanism to do this. It is great that Discovery uses a per-experiment unique user token to do user-level grouping. The system that fundraising uses does not do this, leading to many false positive test results.

Sep 20 2016, 7:04 AM · Analytics-Radar, SRE, Traffic

Aug 30 2016

ellery added a comment to T139790: Log the title the user wants to give the new article in the target language in "Create from scratch".

@schana Has this been deployed yet?

Aug 30 2016, 5:11 PM · GapFinder
ellery added a comment to T142009: Related Pages recommendations user study design.

One idea for comparing the two would be to run an quick experiment on Amazon Mechanical Turk. For some set of articles, generate recommendation sets from both systems and ask the turkers to compare the quality/relevance. You would have to take some care in designing your question, but it could be something along the lines of: "Which set of recommended topics would you consider more relevant to the seed topic?" Then we can get a confidence interval over what fraction of users prefer one version over the other. After choosing a set of seed articles and nailing down the question, this should be a pretty fast and cheap way to get a first assessment.

Aug 30 2016, 5:03 PM · Reading Epics (AI Based Reading Recommendations), Design-Research, Reading-UX-Research
ellery claimed T144286: Sync with CX team on use of recommendations.
Aug 30 2016, 4:51 PM · GapFinder
ellery created T144286: Sync with CX team on use of recommendations.
Aug 30 2016, 4:51 PM · GapFinder
ellery renamed T144285: Measure Impact of Recommendations on CX from Measure Impact on Recommendations of CX to Measure Impact of Recommendations on CX.
Aug 30 2016, 4:49 PM · GapFinder
ellery renamed T144285: Measure Impact of Recommendations on CX from Measure Impact of Recommendations of CX to Measure Impact on Recommendations of CX.
Aug 30 2016, 4:49 PM · GapFinder
ellery added a comment to T144285: Measure Impact of Recommendations on CX.

It seems like CX logs a special campaign name for translations started from the suggestions pane:

Aug 30 2016, 4:48 PM · GapFinder
ellery moved T144285: Measure Impact of Recommendations on CX from Backlog to in-progress on the GapFinder board.
Aug 30 2016, 4:47 PM · GapFinder
ellery created T144285: Measure Impact of Recommendations on CX.
Aug 30 2016, 4:47 PM · GapFinder
ellery created T144284: [Placeholder] GapFinder Continued Engagement.
Aug 30 2016, 4:39 PM · Epic, GapFinder
ellery created T144283: GapFinder Announcement.
Aug 30 2016, 4:35 PM · Epic, GapFinder

Aug 29 2016

ellery added a project to T127531: Harassment and User Retention: Epic.
Aug 29 2016, 8:09 PM · Epic, Discussion-modeling (of Toxicity)
ellery renamed T127531: Harassment and User Retention from [Placeholder] Abuse and Retention to Harassment and User Retention.
Aug 29 2016, 8:08 PM · Epic, Discussion-modeling (of Toxicity)
ellery updated subscribers of T144234: Detox notebooks on PAWS.
Aug 29 2016, 7:41 PM · PAWS, Discussion-modeling (of Toxicity)
ellery created T144234: Detox notebooks on PAWS.
Aug 29 2016, 7:41 PM · PAWS, Discussion-modeling (of Toxicity)
ellery moved T144065: Put EDP models on Demo from Staged to Backlog on the Discussion-modeling (of Toxicity) board.
Aug 29 2016, 7:37 PM · Discussion-modeling (of Toxicity)
ellery moved T139704: Discussion modeling - release notebooks; write up and present results from Backlog to In progress on the Discussion-modeling (of Toxicity) board.
Aug 29 2016, 7:37 PM · Epic, Research-and-Data-2016-17-Q1, Discussion-modeling (of Toxicity), Research-Freezer
ellery added a comment to T139704: Discussion modeling - release notebooks; write up and present results.

The paper draft is in progress and is unlikely to be fully complete by the end of the quarter. The submission deadline is October 24.

Aug 29 2016, 7:36 PM · Epic, Research-and-Data-2016-17-Q1, Discussion-modeling (of Toxicity), Research-Freezer
ellery added a comment to T139704: Discussion modeling - release notebooks; write up and present results.

We presented at the research showcase, at monthly metrics, and the Online Harassment Workshop at the MIT Media Lab.

Aug 29 2016, 7:34 PM · Epic, Research-and-Data-2016-17-Q1, Discussion-modeling (of Toxicity), Research-Freezer
ellery added a comment to T139704: Discussion modeling - release notebooks; write up and present results.

Code is up on github. Once we publish the data and paper, we will want to put some of the notebooks on PAWS.

Aug 29 2016, 7:31 PM · Epic, Research-and-Data-2016-17-Q1, Discussion-modeling (of Toxicity), Research-Freezer
ellery updated the task description for T139704: Discussion modeling - release notebooks; write up and present results.
Aug 29 2016, 7:29 PM · Epic, Research-and-Data-2016-17-Q1, Discussion-modeling (of Toxicity), Research-Freezer
ellery moved T139703: Design and evaluate ''attack'' and ''aggressiveness'' models on article talk comments from Backlog to Done on the Discussion-modeling (of Toxicity) board.
Aug 29 2016, 7:25 PM · Epic, Research-and-Data-2016-17-Q1, Discussion-modeling (of Toxicity), Research-Freezer
ellery renamed T127533: Detox Data Release from Publish and Document User Talk Diff Dataset to Detox Data Release.
Aug 29 2016, 7:24 PM · Epic, Documentation, Discussion-modeling (of Toxicity)
ellery moved T127533: Detox Data Release from Staged to Backlog on the Discussion-modeling (of Toxicity) board.
Aug 29 2016, 7:12 PM · Epic, Documentation, Discussion-modeling (of Toxicity)
ellery added a comment to T139703: Design and evaluate ''attack'' and ''aggressiveness'' models on article talk comments.

We gathered labels for 50k article talk pages and built models that generalize to both the user and article talk namespaces. The following sample file shows ROC scores on held out data broken down by namespace.

Aug 29 2016, 7:12 PM · Epic, Research-and-Data-2016-17-Q1, Discussion-modeling (of Toxicity), Research-Freezer

Aug 27 2016

ellery added a comment to T143702: Eval on NDA data.

See https://github.com/ewulczyn/wiki-detox/blob/master/src/modeling/eval_nda_data.ipynb

Aug 27 2016, 8:43 PM · Discussion-modeling (of Toxicity)
ellery moved T143702: Eval on NDA data from Staged to Done on the Discussion-modeling (of Toxicity) board.
Aug 27 2016, 8:43 PM · Discussion-modeling (of Toxicity)