There seems to be a problem with the output files in hebrew. There are
reports (see below) that the information is coming out as gibberish. It
sounds like Wikimetrics is accepting the usernames fine, but the output is
bad. Might it be the browser or is this an internal problem? They did not
specify which output (CSV or JSON) is having the problem.
Michal and Shai - you need to be part of the mailing list in order to
reply with any more info.
---------- Forwarded message ----------
From: שי כץ <shai(a)wikimedia.org.il>
Date: Thu, Nov 12, 2015 at 2:51 AM
Subject: Re: Problem with user names in Hebrew - metrics
To: Tighe Flanagan <tflanagan(a)wikimedia.org>
Cc: Michal Lester <mlester(a)wikimedia.org.il>, Edward Galvez <
I spoke with Chen on the subject and that's what I understand:
The problem is in the output file. There's no problem with the cohort as a
whole, but only with extracting data on specific users whose names are
written in Hebrew.
We were able to extract data for all the specific users, but in the user
name column, names in Hebrew appear in gibberish. As a result, in the case
of users whose names are written in Hebrew, we cannot link the numerical
data to the specific users to which it relates.
Tighe, does it happen also in the case of Arabic?
Thanks a lot,
2015-11-09 17:23 GMT+02:00 שי כץ <shai(a)wikimedia.org.il>:
> Hi Tighe,
> Thanks for the quick and helpful response.
> Our metrics expert is Chen Spector. She is supposed to be in the office on
> Wednesday/ Thursday. We will consult her and then will be able to
> reply better.
> Greetings from Israel,
> 2015-11-09 16:34 GMT+02:00 Tighe Flanagan <tflanagan(a)wikimedia.org>:
>> Hi Michal,
>> I just did a quick test with some cohorts with Arabic script and it's
>> working in that case. Can you tell us a bit more about the steps you're
>> taking and where it's breaking? Is it a new cohort? Are the usernames
>> validating properly?
>> Are you uploading a CSV or pasting in usernames in the text box? I've
>> seen people have issues with the upload if the file has the wrong character
>> set/language encryption in the file itself. Make sure it's Unicode UTF-8.
>> Alternatively, I've had most success copying usernames from the source file
>> and pasting into the text box when uploading a new cohort. It removes the
>> character set issue.
>> If this is from an existing cohort and the results are illegible because
>> of character coding issues, I'd reach out to the Analytics mailing list
>> (listed on the support page of the Metrics tool
>> <https://metrics.wmflabs.org/support>). They're quite responsive, and
>> perhaps there's an issue they've been working on that has impacted your
>> results unknowingly, and they'll want to know exactly what you're doing and
>> what's breaking so they can troubleshoot.
>> Hope this helps!
> Tighe Flanagan
> Manager, Wikipedia Education Program
> Wikimedia Foundation
> On Mon, Nov 9, 2015 at 6:26 AM, Michal Lester <mlester(a)wikimedia.org.il>
>> Hi Edward,
>> I hope this email finds you well.
>> We have problems with user names in Hebrew when using them in the
>> metrics. They became list of symbols. When we what to compare the results
>> to the user name it is impossible.
>> What do you do in other cases?
>> Thank you, Michal
>> *Michal Lester,*
>> *Executive DirectorWikimedia Israel*
>> *http://www.wikimedia.org.il <http://www.wikimedia.org.il/> *
>> *972-50-8996046 ; 972-77-751-6032 *
>> *[image: תמונה מוטבעת 3]*
> *שי כץ *
> *רכזת הדרכה וחינוךויקימדיה ישראל*
> משרד: 077-3200796 | נייד: 054-4435834
> [image: תמונה מוטבעת 3]
*שי כץ *
*רכזת הדרכה וחינוךויקימדיה ישראל*
משרד: 077-3200796 | נייד: 054-4435834
[image: תמונה מוטבעת 3]
Learning & Evaluation