My wife and I spent a good 5-10 minutes after Tuesday's show trying to figure out how Watson came up with Toronto. I run big server iron for the day job, she's an editor and language geek, so between the two of us, we're as good as any amateurs can be at analyzing the faults in a natural-language AI. :)
The best we could come up with was a combination of factors. For the sake of completeness, here's the clue again:
"Its largest airport is named after a World War II hero; its second largest, after a World War II battle."
Watson worked on key words/phrases in the clue. If you make an assumption that repeated words are less likely to be key words (a quick way to eliminate articles, prepositions, and other filler), then the only unique words in this clue are: airport, named, hero, second, battle.
It's only a recent trend to name buildings after people using both their first and last names. Most just have the last name, and that last name could refer to a lot of people without knowing other context. Chicago O'Hare...heck, it wasn't until Tuesday night that I realized that it referred to naval aviator Butch O'Hare. (I did guess the clue before time was up.) Watson's word association database wouldn't have had that kind of context available, probably.
Without going into an extensive search of airports in North America (or the world) named after people, we did see that Toronto has an airport unequivocally named after a "battle hero". (That being Billy Bishop). But without other stuff in the clue to work with, that led to a lot of uncertainty. Hence five question marks.