We don't "reduce" the collision. We have a bunch of collisions, a lot of background and some signal hidden in there. We need to select some of the collisions according to some properties so that the signal selection is close to 1, and the background one as small as possible. A a result, after selection, the signal over background ratio is more acceptable so that we can study the signal.
Is it clearer?
I think so.....so basically we are trying to isolate the collisions of interest while keeping those we are not interested in to a minimum.
Fascinating.....thanks for the explanation.
Yes exactly. For instance, if one is are looking for the Higgs boson in a final state where there are four muons, we will ignore, in our analysis, any collision that does not contain any muon. This sounds logical, but the basic idea is there :)