Data Mining Research (DMR): Who are you and what is your background?
Roberto Battiti (RB): I am full professor of Computer Science at Trento University, received the Laurea degree in Physics from the University of Trento, Italy, in 1985 and the Ph.D. in Computation and Neural Systems from the California Institute of Technology (Caltech), USA in 1990. My main research interests are heuristic methods for problem-solving, in particular Reactive Search Optimization, which aims at embodying solvers with internal machine learning techniques, data mining and visualization. I am a Fellow of the IEEE, for contributions to machine learning techniques for intelligent optimization and neural networks, author of highly cited publications, and one of the co-founders of the startup Reactive Search SrL
DMR: How did you come up with the idea of your book “Reactive Business Intelligence”?
RB: It grew out from my passion (shared with my colleague Mauro Brunato) for combining business intelligence with learning components. Building learning systems, following the human paradigm of “learning on the job” is my old obsession, leading to the introduction of Reactive Search Optimization ideas in the last decades.
DMR: Why this new term? What does it mean to you?
RB: The word reactive hints at a ready response to events during the business intelligence process through an internal online feedback loop for the self-tuning of the system.
Developing reactive models which rapidly adapt to the decision maker wildest desires, this is our objective.
A connection between visualization and problem-solving strategies is also at the heart of RBI: the decision maker can be, and should be, in an interactive loop, rapidly reacting to first results and visualizations to direct the subsequent efforts to suit his needs and preferences.
DMR: What advice would you give to people involved in data mining / business intelligence?
RB: To forget about computers to concentrate their creative energies to navigate from data to insight. I am exaggerating a bit of course, but computers are now so powerful that they should become invisible to the user. To make a concrete example, the user should not be asked to read manuals to start analyzing his data. This is why we think that having the system learning the user preferences can help. This is a long-term goal of course, but the path is clear.
For more information, you can try their tool Grapheur.
