A note worthy initiative to capture this valuable information. Another one that may be worth looking at is http://internetdealbook.com/category/top-deals/. It focuses on the more visible deals and does include deal size where public.
It looks like it's just scraping the portfolio pages of most of these, and most portfolio directories list nothing more than a company's name and website.
That said, doesn't AngelList or CrunchBase have something like this already? (or at least an API to extract more information from)
Now, it would be fun to run predictions on the portofolio and investment data. Predict what companies will get investment next, what domains are hot, and so on. Who knows, you could even make money from this.
/e math isn't far away, but it helps to put it in context. On that note, it's nice to have an updated list with most vc deals made in the past 3 years