• Period: 2006-2008
  • Funded by:
  • Partners: University of New Caledonia and University of Lyon.
Easy prototyping of data mining algorithms (iZi lib)

Although guided by real problems, data mining techniques are still marginally used and their implementation can only be carried out by specialists programmers familiar with low-level code. The technology transfer is thus slowed down by some limitations, among which the time necessary to the development of operational programs. We consider here an important and broad class of data mining problems: the discovery of interesting patterns in large databases. From a theoretical point of view, these problems must be representable as sets. In this setting, we propose a library facilitating the resolution of such problems, based on the use of generic and scalable algorithms. The characteristics and optimizations of the algorithms are transparent for the programmer, and only the development of the specific properties of its problem is left to him/her. The library, called iZi and developed in C++, is available under GNU General Public License. For more details, see the technical documentation.