• data mining
• patent
• patent analysis
• patent mining
• text mining

The aim of this thesis is to create patent analysis supporting instruments. At first, it is led a critical study on prior art about patent mining and information extraction from patent. With the same logic, the treatise shows a new model for research keywords classification and a flexible proceeding for retrieval of interesting patents. Furthermore, the dissertation deals with a study about patent legal claims and an algorithm for the automatic creation of a matrix that relate invention features to patent claims. Finally it is introduced a methodology which was developed in collaboration with CNR of Pisa and Erre Quadro Srl; this methodology aimed to extract information automatically from patent texts.