IMHO, there is something compare the golden output and DUT output in testbench (I call it Checker). But in verification book, there is both Scoreboard and Checker. Are they similar?
No they are not. Checker basically only tests for proper functionality for example you can have a checker for overlooking the protocols for your bus. Checker only checks that if transactions stick to the specification. Scoreboard is basically to collect coverage, if your protocol supports burst-4, burst-8 and single type of transaction , it is the scoreboards job to keep track of how many burst-8, burst-4 and single type transactions were sent. Suppose scoreboard shows that no burst-8 type trans ever took place then you know that u have a hole here and you need to fix this hole. Note here that checker flags for error but all that scoreboard does is to keep score of the things that took place in your simulation.
I found you describe scoreboard like functional coverage
I have read a Mentor AVM document like find scoreboard is something like "reference model + checker". And I think functional coverage should be isolated from the idea of scoreboard, though they are similar.
Checkers are more to check functionality interms of protocol,behaviour of DUT ...
Scoarboard is a collection module , which intern can used for checkers or functional coverage,BFM development ... so as such score board is more like a collection , so it is upto user how he want to use collected data .