Boolean Methods and Logical Analysis of Data

About 20 years ago, we proposed an innovative approach to data mining based on a blend of Boolean techniques and combinatorial optimization. The basic tenets of this approach were presented in a joint paper co-authored with Toshihide Ibaraki and myself. It was subsequently developed by Peter Hammer and his coworkers into a new broad area of research, dubbed Logical Analysis of Data, or LAD for short. The effectiveness of the LAD methodology has been validated by many successful applications to real-life data analysis problems.

In a first part of this lecture, I will propose a brief overview of some fundamental Boolean models and of illustrative applications arising in computer science, in optimization and in game theory. I will next turn to a presentation of the basic principles of LAD and of some of the theoretical questions that have been investigated in connection with this methodology. This part of the lecture should allow the audience to place the development of LAD in a historical perspective.