An evolutionary classification tool for ViaLactea, will catalogue clumps in terms of the evolutionary stage and mass regime of the ongoing star formation. There are two components that need to be developed at the foundation of the classification tool:

• an evolutionary classification toolbox
• a set of star-forming clumps in known stages of evolution to be used as a training/test-set for machine-learning algorithms...and adopt some kind of evolutionary scheme

In a preliminary view, we foresee two different data mining approaches to source classification:

Supervised Classification

For a subsample of points, its category/class is well known. It is needed an order of 1K of objects to be used as a training set and a balanced population of classes in the training set.

Weak Gated Classification

We know nothing about the sources evolutionary stage;

Identify over-densities in the given parameter space (e.g., built on the evolutionary toolbox, plus any other available evidence). Data are then grouped into clusters, i.e. groups of data entries sharing common but a priori unknown correlations among parameter space features.