Virtual space opens new possibilities of expression and experiment, in addition to the potential for simulation. It also helps to re-create and remould a monumental complex without having all the information and to test hypotheses that we would otherwise be unable to validate without compromising the heritage values of a site by physically reconstructing it. This was the case of the Roman theatre of Byblos (A.D. 218) in Lebanon that, in the 1930s, was moved and rebuilt near the sea by archaeologist M. Dunand. These new possibilities present new challenges to our Computer Aided Design Research Group (GRCAO) of the Université de Montréal and allow us to develop our know-how in the restoration of monuments and archaeological complexes. Our main goal is to define new approaches to modelling and computer-based descriptions that make these intuitive interactions possible and that propose an organization of information that optimizes the understanding of these interactions. Through the study case of Roman theatres and precisely the Roman theatre of Byblos, we intend to propose a codification of the information which is necessary for a better understanding of the site and the monuments preserved in situ by proposing digital models. These models reflect not only the result of the experimentation but far more the process that produced the result.