“Intelligent” Data Mining: theory and applications
What is Data Mining? What kind of information can it extract from large and complex amounts of data? What are the possible applications of this innovative method of mathematical analysis? Massimo Buscema and Giulia Massini, director and senior researcher at Semeion – Research Center of Sciences of Communications in Rome, explain it in a document entitled Introduzione Teorica al Data Mining intelligente, published recently.
The main objective of Data Mining is to extract the most important information, visible or “hidden”, from a data set useful to those who need to make decisions or simply better understand the material under consideration. Hidden information indicates the “weak signals” that may escape traditional statistical analysis. This is why we talk about “Intelligent” Data Mining, achievable through a nonlinear analysis of the same data available.
But what is it meant by nonlinear analysis of data? Researchers at Semeion explain it with an easy example: “From the point of view of linear interpolation, Norway’s rugged coastline would appear as a succession of straight lines: a good way to approximate the overall shape of the region, but really bad to provide a map to those who need to navigate the fjords”. In short, this innovative method, able to “trace” a phenomenon – however complex – more closely, provides more interesting and useful information, especially predictive of future trends of a phenomenon over time, which is impossible with the traditional techniques, however sophisticated they may be.
The document examines with graphs the mathematical aspects of the method, based on artificial neural network systems, ie advanced applications that simulate the way in which the human brain processes information, but with a computing power enhanced by latest generation hardware and software. Moreover, the effectiveness of these innovative artificial intelligence algorithms is demonstrated with a real case: the optimization of the management of the Supplementary Health Fund of the chemical-pharmaceutical Collective Labour Agreement (Faschim), a research project recently presented during a conference held at the Ministry of Health in Rome.
Put to the test, the system was able to identify typical and atypical behaviours of the fund beneficiaries, as well as potential “hidden” frauds in the data, to the advantage of both the users and the economic and social system that helps improve public health.