Thesis of Olivier Cavadenti

Manufactured product trace mining

Defense date: 27/09/2016

Advisor: Jean-Francois Boulicaut


In the context of manufacturing, traceability unit should provide answers to many questions about the life cycle of each artifact identified and it addresses many issues of safety and risk management. The authentication strategy of each object depend on the capacity of the system under developpement to capture the obligatory steps of the path by recording and analyzing the essential steps.
Authentication that interests us is to ensure the authenticity of the objet across all transformations, and so flow wich allowed to make, constitute and make available it to a given place at a given time. The appoach considered must take into account very large volumes of spatio-temporal data. These data can be called traces et the characteristic of traceability unit will feature a large collections of traces with specific intrinsic structures.
The aim of this PhD is to enhance the collections of unit traces offering traces mining functionality (discovery of regularities and irregularities, interesting episodes, signature of targeted behaviors,etc.). For this,it is planned to propose language of patterns and to develop new tools of constraint-based data mining to assist this enhancement.