Julio Valdes talked about data mining. Which is, apparently, "the non-trival process of identifying valid, novel, potentially useful, and ultimately understandable". The goal of his research is to combine the best results from the human and the machine in a visual tool.
Why use a visual representation at all? He argued that the amount of information in relational structures is large, humans are well qualified to understand complex geometric structures. Additionally, when you present results visually you need little mathematical knowledge to understand the results.
He talked at length about the formal mathematics behind representing data, and then demonstrated some "virtual spaces" for data representation.