Freebase has
40,760 WordNet synsets mapped to equivalent Freebase topics and another 2,300 mapped to broader topics. This could be used to bootstrap any efforts in this space. Of course, it's out of a total of almost 118,000 synsets, so it's only 35-40% coverage, but it probably covers the most popular/common topics. Something to be careful of is semantic drift in Wikipedia/Wikidata, since these alignments were originally done back in 2009.
Here's the
schema for the types associated with Wordnet in Freebase. (Yes, I know types are evil and illegal in Wikidata, but just consider them collections of properties grouped together for human organizational convenience.)
BTW, I agree with Daniel that care is needed here. It's easy for casual observers to get confused about what should be connected, particularly since Wikipedia articles aren't necessarily divided along strict semantic lines.
Tom