I cooked up a prototype of this project about a year and a half ago. While the results were encouraging, I lost motivation and it just collected dust for all that time. In the past 3 weeks I've revived the project and given new life to it.

The project started when I wondered: "What if delicious tag signatures were vectors, would their nearest neighbors in n-dimensional space be a good match for similarity?" The answer took awhile to figure out, but it was a yes. For me at least, it wasn't an easy problem to solve; finding the nearest neighbors with each vector having 10 dimensions, with total dimensional space about 50,000. Figuring out a way to solve this in a timely manner was a joy.

At the time this project was first started and the first prototype completed, there were absolutely zero sites, other than google's sub-par "similar" link, that offered this. Well, there was "similicious" but its index was tiny, and results weren't too impressive. Now I count 5 full fledged services that do a pretty good job. Some are ranked quit highly and probably make a decent buck. I really think I missed a decent opportunity by not pushing through the last 20% of the project, and losing motivation entirely.

But, like they say, better late than never.

Questions/comments are welcomed/anxiously awaited.