Question regarding the sampling algorithm and scoring function of HADDOCK

When I read about autodock4 for example it said that it primarily uses Lamarckian genetic algorithm.

However, with HADDOCK I haven´t found any article describing the sampling algorithm or scoring function concretely. I know that the AIRs are used to create the interface. But how does HADDOCK sample configurations within this interface and which scoring function does it use to rank these? Are there any article about HADDOCK that answers my question?

Have you never taken a look at the HADDOCK software page?!

Check: https://www.bonvinlab.org/software/haddock2.2/

The 2003 JACS paper is the original HADDOCK publication describing the protocol.

Many more papers followed that one.