Mineração de textos de patentes para monitoramento tecnológico, identificação de aplicações e seleção de materiais: estudo de caso em desenvolvimentos de aços

Abstract

The growing number of options in materials leads to increasing complexity of the process to select the most suitable to perform a function in a product or to choose a competitive application for a newly developed material. Computational resources and databases are used to minimize uncertainties, but they still lack alternatives to evaluate opportunities that may not present themselves as commercial options. The bibliographic records of patent documents are presented as comprehensive and updated information sources, with technical details that can be retrieved from bibliometrics and text mining techniques. This thesis aimed to develop a data sheet based on text mining in steels patent records indexed in the Derwent Innovations Index. The methods employed made it possible to retrieve 85% of text segments among 22,911 records between 2009 and 2018. It was possible to evaluate recent steel developments based on chemical composition, processes, microstructure, properties, and applications. The results show the progress in the development of more detailed indicators for technological monitoring activities. In the scope of materials selection, it was possible to create diagrams of materials properties over time and other information retrieved. The procedures used in this study can be improved for a longer time scope, other properties analysis, or application to different classes of materials.