Patent Mining to Understand Functional Evolution of Engineered Products

Abstract

Abstract This work presents a patent mining methodology to qualitatively examine functional evolution of engineered products. Using a topic modeling method called Nonnegative Matrix Factorization, topics of nutcracker- and drone-related patents of different time frames are automatically generated to understand how the two products have functionally evolved over time. The topic modeling results demonstrate that nutcracker’s nut cracking mechanism has changed from an impact method to a rotating method for facilitating mass production. Drone system has evolved from a mechanical system with flight capability to an electromechanical system widening its application in package delivery and videography. The topic modeling results provide different design insights that could assist designers during the early product development process. In addition to the topic modeling approach, an analysis with patent citations is performed to gain additional insights about the functional evolution of the engineered products. The patent citation analysis suggests that the topic modeling approach provides a reasonable illustration of how products have evolved over time, but the algorithm requires improvement as some functional aspects of a product can be obscured by topics that simply characterize the technology. The objective of the study is to provide a basis for automated modeling of patents to understand the functional evolution of engineered products.