Optimisation génétique et interactive de sites web

We deal in this paper with the problem of automatically generating the style and layout of a web site in a real world application where thousands of web sites are considered. We review the main difficulties of the problem like taking into account the user aesthetic preferences. We propose the use of an interactive genetic algorithm to generate solutions that satisfy user preferences. This work represents as far as we know one of the first real world application of interactive GAs. We have defined two encodings, one for the style and one for the layout. We show typical results obtained with our system and how user interact with it. We conclude on this work by analyzing further developments of this research in other domains.