So we came upon this new restaurant website called Foodio54 (no, wait, we were emailed about it. Getting stuff sent to you is the new browsing), and we want share. We don’t really want to since it’s a competitor of sorts, but it’s more in the same genus rather than the same species, and somewhat inherently interesting. Basically, the site purports to be an intelligent restaurant suggester, and that intelligence is based on comparing the restaurants that a user rates highly to the favored restaurants of other users who share that user’s tastes, and coming up with a list of restaurants that the user ought to like based on algorithmic analysis of the database. They call it “collaborative filtering,” and it’s also used by large commercial sites like Netflix and Amazon.
The reason why collaborative filtering works on the aforementioned sites is that they have a huge database of user preferences to mine in order to generate their recommendations. For Foodio54 to produce relevant results, it will need a similarly large dataset, or else the available information will be too sparse to be actually helpful. Getting people to rate movies on Netflix or products on Amazon is easy enough, since they’re already there, making purchases. Getting a critical mass on Foodio54 to rate enough restaurants for collaborative filtering to work well will be a challenge, especially given the upwards of half a million restaurants in the database and the fact that no one’s actually buying anything.
Meanwhile, the two applets in the “Spotlight Features” caught our attention for being really neat and well executed. The first provides a graph of Starbucks saturation by distance from any given zip code (and compares it with state and national averages!), and the second is a 2008 campaign tracker that gives the location (and nearby restaurants) of upcoming campaign stops for each of the nineteen wannabes, including Fred Thompson. Now that’s information we can use.
Foodio54 [Official Site]