Hello Emmanuel, it's not so much a translation software (like Google Translate), there is no way to translate sentences or parts of sentences, but you can look up single word translations (or whatever the stored unit on Wiktionary is). Its main purpose is to make a subset of Wiktionary available offline (and adapt it for mobile consumption). The iOS app itself is not open source, but big parts of the extraction software are. the main work is done by JWKTL (Java-based Wiktionary Library, https://github.com/dkpro/dkpro-jwktl), where I'm also a contributor. JWKTL parses the data into a BerkeleyDB file; I wrote a small adapter to use SQLite instead. It should be easy to do something similar for the ZIM format. However, more recently I've started to move away from JWKTL since a some data cannot be extracted from the raw mediawiki markup alone. A good example are pronunciations, which are often entirely generated in Lua. For this I use a combination of microformats2 and a simple Python parser. See this thread on phabricator for more details: https://phabricator.wikimedia.org/T138709#3048681 This often requires some minor changes to templates or modules. My plan is to add special markup for the most important features, which should also help with the transition to Wikidata. Jan