Well done for the engineering feat here. Parsing Wikipedia at scale isn't easy.
Very surprised that they don't mention DBpedia though (http://wiki.dbpedia.org/) or discuss how they navigate through some big NLP challenges with Wikipedia, namely that many edits are made by bots.
Calculating context on mobile is a huge win for content providers, who can use the context to create deep links into other apps.
Example: I'm reading an article about James Bond, and ideally the phone can calculate that there's a new Bond movie, and I'm near a theater, and I'm free on Thursday night, and then create a link to buy tickets.
Very surprised that they don't mention DBpedia though (http://wiki.dbpedia.org/) or discuss how they navigate through some big NLP challenges with Wikipedia, namely that many edits are made by bots.