Why the reconstructed jet need to substract the "background"? There is a trouble bother me a long time. I know it may be stupid 
question. But I really need your help. It's in the CMS paper:
CMS Collaboration, CMS Physics Analysis Summary No. CMS-PAS-HIN-12-013, 2012,
visit http://cds.cern.ch/record/1472734
In the section of jet reconstruction, it already contain the jet selection process when reconstruct the jet. More specifically, in this case, the minimal transverse momentum of all charged particles was set to 1 GeV/c. And the transverse momentum of a jet is required to be larger than 100 GeV/c. There are also restriction on $|\eta|$.
So, after the selection process, the constitutions of the reconstructed jet are the particles that we need. 
Why we still need to substruct the background from original reconstructed jet?  After the selection, why still exist the underlying event or what are the constitutions of the background?
 A: 
In the section of jet reconstruction, it already contain the jet selection process when reconstruct the jet. More specifically, in this case, the minimal transverse momentum of all charged particles was set to 1 GeV/c. And the transverse momentum of a jet is required to be larger than 100 GeV/c. There are also restriction on |η|.

This is the way for trying to catch in a wide net possible jets. A lot of background is rejected by these cuts, presumably cuts  chosen using monte carlo data. In a sense it is the difference between necessary and sufficient for eliminating background. It is necessary to have a rough cut, so as to work with an enriched in jets sample. Then one has to use detailed monte carlo simulations to estimate the background remaining in the sample and define corrections to it. 
The paper is complicated and needs serious study to get the details. The background still remaining after the first selection is the same as the one rejected by the first cuts, that remains within the kinematic region accepted for the jets, overlapping them, i.e.:

standard selection criteria are applied to remove detector noise background, beam gas, beam scraping and ultra-peripheral-collision events from the selected events . 

There seem also to be extra criteria for heavy ion on heavy ion collisions, but the logic is the same : general cut to select enriched sample, and background remaining in sample calculated  by monte carlo and symmetry methods.
