Poking around in those datasets is fun. That reminds me of a similar thing I did a few months ago with the same database (<a href="http://calori.org" rel="nofollow">http://calori.org</a>) as part of a tech demo/test.<p><pre><code> There's a lot of data included which, quite frankly, doesn't
seem very useful (e.g. nutrient "22:1 undifferentiated",
whatever that is.)
</code></pre>
One thing that struck me as well was the amount of cleanup necessary to bring this data into a somewhat presentable shape. I guess is that at some point they just dumped their messy Access DB into CSV and called it a day. But it's still great that they did decide to make this public domain.