Recommendation Engine - handling thumbs up/down

Hi all;

First off, for anyone else reading this, start with this post - the answer there is incredibly helpful.

Ok, on to the question. My use case is a bit different but we'll do a concert website. The user has gone to a number of concerts. We can look at the ones they went to, find upcoming ones that match, all using vectors and nearest neighbor to find good matches.

But, I give the user the ability to give a thumbs up or thumbs down to the concerts they went to. I obviously don't want to make the rating part of a vector. I also worry that a single rating has to encapsulate everything about a concert. The artist had a lousy night so thumbs down yet the type of music, venue, etc. would normally get a thumbs up and so I don't want to make this a hard inclusive/exclusive pair of lists.

So what is normally done here? Would a thumbs up/down be used as a multiplier to decrease/increase the nearest neighbor distance before selecting the best matches? Or is there another approach?

And similar issue, I also look for other users where I do a nearest neighbor search on the user vectors. And then find the concerts that user has attended. And again, they have provided a thumbs up/down to those concerts. From that I pull the list of upcoming concerts that user is going to as possible suggestions. How do I take into account the thumbs up/down that other user set to their past concerts?

thanks - dave