[ostegm]

Can you describe a little about how you go from raw text to a summary? Is this supervised learning? If so, what is the training data?