Patent text mining and informetric-based patent technology morphological analysis: an empirical study

Abstract

Patent technology morphological analysis utilises the advantages of both patent information analysis and morphological analysis to provide more detailed information on current and future patent technology. Current patent technology morphological analysis methods are largely reliant on manual expertise in the construction of morphological boxes with few approaches to the evaluation of future morphological configurations. We developed a patent text mining and informetric-based patent technology morphological analysis technique. We defined the basic parameters of the morphological box as the factors in factor analysis, with a patent keywords matrix, and the clusters in clustering analysis with factor scores. Patent citation, year of patent registration, keyword frequency, and contributing factors were used to evaluate future morphological configurations. We used an empirical study of liquid crystal display wide viewing angle patents to validate our method. The Thomson Reuters Derwent Innovation Index was used to collect patent text datasets. Our study indicates that this method is feasible for the implementation of patent technology morphological analysis. Our method provides advantages in terms of cost and time reductions during morphological box construction and more flexible methods for evaluating morphological configurations. We discuss problems and possible development of our method in the last section.