Hello,

We’ve published a series of thematic subsets based on Wikipedia. The idea sort of follows what we had been doing for WikiMed[1], e.g. provide a single, topic-specific zim file based off the same Wikipedia articles - so that if you are a physician in an offline setting you needn’t download the full 5,000,000 articles in order to get the 50,000 you were really interested in (and also: 1 Gb download at most vs. 80Gb).

The list is set to expand, but we’re starting with the following:
Football (soccer)
Basketball
Cricket
History 
Geography
Maths
Physics
Chemistry
Comics

The initial selections are based off enwp’s wikiprojects. Most of these are/will be available also in the largest languages (EN/ES/PT/AR/FR/DE/ZH/HI/TR/RU), with some variations as we’re not entirely sure there’s much demand (or material) for Russian articles on cricket. 

These selections are easier to find from the app’s library, but they can also be downloaded directly from our public repo [2]. Now that MWoffliner and ZIMfarm are up and (mostly) running, expect updates to run automatically every month or so. And for those asking, the MWoffliner 1.9 release you saw earlier should allow us to finally update larger zim files (>500k), or so we hope/pray/want to believe.

The other thing you will notice is that the landing page is not your usual « Welcome to Wikipedia » blurb. In fact, we did away with most of the text so that we wouldn’t have to design landing pages in languages we don’t understand. A Very Elegant solution was found by the Most Excellent Joseph Reeve : present it as tiles, with the images directly taken from the corresponding Wikidata entry (P18). The 100 articles constituting this new landing page are themselves picked based on a scorecard that ranks articles based on traffic, evaluation and God knows what else, so that in the end the end user is likely to be presented with a landing page on subject s/he’s likely to be interested in[3]. This image [4] shows the Arabic chemistry, Hindi cricket and English basketball selections’ respective landing pages.

Needless to say, we’ve kicked any Wikipedia design custom and convention squarely in the teeth, and have no regrets about it (but fear not: articles themselves remain unchanged; we’re revolutionaries but not punks). As indicated above, the central reason for this approach is that through wikidata we get the appropriate spelling/name for the articles, and can therefore present users with not only the proper content, but also quite simply the right alphabet. 

Last but not least, the zim files will come in three flavours: Mini (only into and infobox), No pictures, and Maxi (full content, no videos), the idea being that we provide content best suited to users’ needs, bandwidth and storage space.

If you think there are specific topics of interest we should cover -> https://github.com/openzim/zim-requests/issues

Best regards,
Stephane

Kiwix 
Internet content for people without internet access 
www.kiwix.org


1- https://play.google.com/store/apps/details?id=org.kiwix.kiwixcustomwikimed
2- http://download.kiwix.org/zim/wikipedia/
3- *cough* unless you check the football selection *cough* 
4- https://commons.wikimedia.org/wiki/File:Kiwix_selections.jpg