Abstract : Today, a large community of scientists is working to make possible the achievement of a quantum computer, a machine that can offer at least in theory (and especially for problems whose complexity grows exponentially with the size of the system) a degree of performance inaccessible to its classical counterpart. This thesis is looking at the possibility of producing a semi-classical approach of quantum information in two areas of interest: the cloning of a qubit, and the amplification of spin in spin chains. In the first part of this thesis is studied the role of interference in quantum cloners. We study in particular the case of cloners without interference (as defined, in the thesis) that turned out to be an intermediary case (that can be qualified of semi-classical) between purely quantum cloners (which propagate coherences and probabilities of density matrices) and classical cloners (which carry only the probabilities). In the second part, the phenomenon of amplification is studied in spin chains, which allows to amplify the state of a unique spin in a state of polarization of the entire chain, problem for which the semi-classical approach (valid because of the large number of spins) is used to show the unexpectedly important role played by the edge effects in these kind of systems.