Founded by Flickr's Caterina Fake and a team of MIT graduates, Hunch was designed as a collective intelligence decision-making system that predicts what you will enjoy based on past decisions. As the site learns more about you, it's recommendations get stronger. And it has something to say about everything with broad categories ranging from home to art to music.
In order to make the best recommendations, Hunch relies on an ever-growing database of connections -- between you and your friends, between you and your interests, etc. -- known internally as the taste graph. Column Five Media put together an infographic that shows that graph in more detail: "what it takes to produce it, what goes into it, and how it works," Hunch explained on its official website.
Infographics are always a bit of a hodgepodge of statistics culled from a variety of sources. Here, we sort through the clutter and pull out some of our favorite facts and figures:
- 27,000 lines of code went into making the taste graph, which is built using proprietary software written in C and Assembly.
- Hunch runs the taste graph on a supercomputer with 48 processors and one terabyte of memory.
- Despite massive computing power, it takes two days for the taste graph to completely refresh. It currently holds 500 million people, 200 million items and 30 billion edges, or links between users and other users or items.
- More than 80 million THAYs (Teach Hunch About You) questions have been answered and are documented in the taste graph. The average Hunch user has answered 113 THAYs.
- The taste graph includes two million venues that were added by two million people who checked-in to the various locations.
Check out more Infographics on the Technology Channel.
This article available online at: