The organization of the paper is the following...
In section~\ref{XP:Classification}, we aim at classifying every element from a BIM dataset.
We take BIM models from industrial projects and classify their components according to their profiles of engineering.
The idea is to investigate the possibility to sort BIM components using machine learning, and analyse which features are the most relevant.
Then, the section~\ref{XP:Saliency} proposes a method to adapt the visualisation of the model regarding the user profile.
The methods creates an novel saliency model for architecture, and gets a map of the visual attention potential of the model.
From this map, the system slightly varies the colors of elements to attract the attention of the user on the relevant components for his profile.
Figure~\ref{Fig:Process} illustrate the 2 steps for the adaptation of the BIM model.
On the Left, the raw BIM model.
Firstly, we classify the element to create groups in relation to the different profile of engineer in the construction project.
Secondly, from this classified model, we slightly adapt the color to put the visual attention on a specific group of element.