This project reminds of what Wikidata does: organize facts as semantic triples [1] using inheritance in the way OOP does.
Example about our planet: https://www.wikidata.org/wiki/Q2
Example about our planet: https://www.wikidata.org/wiki/Q2
Now I'm thinking the two things are more complementary.
This project seems like a domain specific web scraper, targeted at common usages found on Wikipedia.
There's so much data in Wikipedia that it's a good target for this.
So one combined use would be scrape Wikipedia to find mentions of e.g. tour de France cyclists and then ensure that they are correctly tagged in wikidata.
The author talks more about wikidata in earlier post in the series about this.
This is pretty much what DbPedia does, FWIW. But Wikidata has the benefit of being built for the actual purpose of hosting structured data, regardless of language.