Logical Analysis of Data (LAD) is a novel approach to extract knowledge from data using logical concepts. In its original form, the complete LAD process contains several bottlenecks which make the method not suitable to large scale problems. This project focuses on the elaboration of new algorithms for various phases of the process, in order to reduce the overall computational complexity, so that LAD can be applied to a larger family of problems. Another important goal of our research is to extend LAD to handle classification problems of arbitrary number of classes (so far it is limited to the discrimination between two classes only). A preliminary study of this problem was presented at ISMP'97 in Lausanne.