Whether it be for mining rare objects or performing statistical analysis,
astronomers rely on cross-matches to build mutli-wavelength datasets.
With an increasing number of large and high quality surveys, the need
to deal with more than two catalogues and to apply machine learning techniques
becomes more and more pressing.
After a quick review of cross-matching tools, we will focus on the challenges
of the multi-catalogue algorithms, and provide feedbacks on the development
of the ARCHES project dedicated tool.
We will then talk about automated classification and show how it can be used
to improve purely positional probabilities of identification as well as to look
for rare ojects.