In case you didn't know it, we hosted a session at Wikimania to talk about
access as it is impacted by the cost of data. Offline access is one of the
potential solutions to bridge that gap.
Jorge Vargas led the panel, which I was on alongside Stephane
Coillet-Matillon (Kiwix), Adam Holt (Internet in a Box), Florence Devuoard
(Wikifundi + Wiki Loves Women + more), and James Heilman (Wikimedicine
Project + Wikimedia Foundation board).
We had a great discussion and it was, for me, a great kickoff into the
off.network hackathon the following day.
I've published notes from the session here for those of you who are
As always, let me/us know if you have any questions or follow up comments.
I look forward to continued collaboration in the future!
*Anne Gomez* // Senior Program Manager, New Readers
*Imagine a world in which every single human being can freely share in the
sum of all knowledge. That's our commitment. Donate
A quick note to announce that Kiwix released this morning a new version of the Wikimed App called Wikimed Mini . It basically complements the earlier app in that it is 90% smaller and should therefore make it easier to download and store by users who need it most (50% of installs for the English version happen on the Indian subcontinent; for the French one, 80% are in Africa).
But apart from this PSA I wanted to share the thinking that went behind the design of this particular app, as I’ve been told it could be of interest to this list:
There basically were three reasons for this "mini" version:
1. When connectivity is an issue, size does matter. If you look at the 20+ main offline medical apps out there, you'll see that they all range in the 25-40Mb (the largest being 120Mb). Several of them have 1M+ downloads, and even if I am no physician I'd say their content is rather minimal and that WP content is far better: yet people download the other, smaller apps rather than our gigantic, all-encompassing 1.2 Gb one;
2. You may have heard that thing about 60% of mobile readers not going past the Lead section . We did too, and took the drastic step of removing everything below that - basically keeping only the lead part and infobox. That saved us about 60%;
3. Then we looked at what was left and figured that infobox illustrations weren't that helpful: either because they look good but are not very informative, or because if they are in fact informative offline limitations make it impossible to see more than a thumbnail. Don't get me wrong: the ability to see high-res images comes high on the list of requests, but a choice was to be made and that one was a low-hanging fruit. So we removed the illustations as well, and now we're left with a 90% smaller app.
In spite of its size, the current Wikimed is rated higher (4.7) and kept longer (75% retention at D30) that most apps (4.2/20%, respectively on average): we’ll keep you posted on how the new app fares in comparison to that.
 https://play.google.com/store/apps/details?id=org.kiwix.kiwixcustomwikimedm… <https://play.google.com/store/apps/details?id=org.kiwix.kiwixcustomwikimedm…>
 https://meta.wikimedia.org/wiki/Research:Which_parts_of_an_article_do_reade… <https://meta.wikimedia.org/wiki/Research:Which_parts_of_an_article_do_reade…>
I want to announce here the publication of new datasets to easy make
selections of Wikipedia articles. This data can be used by any developer
or tech-friendly guy to create subset of Wikipedia. You can find the
data here: http://download.kiwix.org/wp1/ (or via FTP).
This data repository will be kept up-to-date every month thanks to a few
scripts which are published here:
https://github.com/openzim/wp1_selection_tools. Of course, everything is
For each of the 500.000+ Wikipedias, you can find there TSV tables which
contain usual indicators of importance for each article: like number of
interlanguage links, number of links pointing to an articles, pageviews,
... All gathered in one file. For the Wikipedia in English you will
benefit in addition of the Wikiproject importance/quality evaluations.
If you are really lazy, there is a "score" file which mix all these
indicators to give a unique score number per article. The methodology is
described here https://github.com/openzim/wp1_selection_tools. For
example, if you want tje TOP1000 articles of Wikipedia, just take the
first thousand lines of the "score" file to get your list of articles.
All this work has been done to allow the creation of TOP Wikipedia
articles ZIM files. It has also been done to make possible the creation
of ZIM extension files, a concept we want to develop to improve our
WikiMed Android apps. Both of them will appear before the end of the year.
Kiwix - Wikipedia Offline & more
* Web: http://www.kiwix.org
* Twitter: https://twitter.com/KiwixOffline
* more: http://www.kiwix.org/wiki/Communication