Research On Tracing Technological Evolution Pathways Based On Patent Documents

With the rapid development of economic globalization and technological development,innovation has become an essential factor for sustainable business development and strengthening an internationally-competitive edge.Technological innovation is a continuous and dynamic process that spans the entire lifecycle of an idea,from scientific research to production.However,improving the competitiveness of technological innovation depends neither on the entire development process nor all its technological components.Rather,it relies on a few key technologies that hold great potential to affect future technological trends.Hence,finding methods to effectively identify and trace the development of these key technologies has become an important topic for researchers and managers in the field of technology management.Focusing on the core issue of identifying the evolutionary pathways of a technology,this thesis presents a systematic framework of bibliometrics,text mining,and social network analysis techniques to help the field of technology management analyze the evolution of technologies by using patent documents as the data source.The framework comprises a theoretical framework with a methodological basis to: study the lifecycle of a technology;trace the transition of technological areas and topics;and explore the key pathways of a technology's development.The main content of this thesis is summarized as follows.First,explaining the theoretical basis of,and motivations behind,technological evolution.After clarifying the related concepts and the significance of technological evolution and its pathways,the four theoretical foundations of technological evolution are introduced: technological paradigm theory,path dependence theory,technological portfolio theory,and technological evolution theory.The dynamics of “demand-pull” and “technology push” are explored in the context of technological development and its related concepts,e.g.,the technology roadmap.Second,outlining a new method for analyzing the lifecycle of a technology based on patent citation networks.The different stages of a technology's lifecycle tend to show different characteristics.Therefore,analyzing these stages is necessary to chart the course of a technology's evolution.Further,patent citation networks reflect the direct and dynamic relationships that have contributed to the formation of a technology.These relationships can be mapped to simulate the interactions and growth of different technologies during the innovation process.Within the method,the growth curves reflected in a citation network are matched to various technological characteristics and distinguished into different stages of a technology's lifecycle.A dynamic citation network is more likely to reflect the characteristics and pathways of real-world technological evolution,and excluding isolated nodes helps to reduce potential misinterpretations caused by noise or redundant data.Third,developing an approach for tracing the evolution of technological areas and topics relies on patent classifications and the content of those patents.Patent classifications are used to construct a co-classification network,which,in turn,can be used to determine the distribution of technological endeavors through a community discovery algorithm.These distributions reveal the emergence,growth,and alternative pathways pursued across different technological areas at different stages.However,a single information source cannot accurately represent the multi-dimensional relationships between patent documents.Therefore,a model based on latent Dirichlet allocation(LDA)is used to explore the relationships in and between the content of the patent documents and the technological topics extracted from their titles.This approach allows analysts to trace the birth,derivations,fusions,and disappearance of technological topics according to topic similarity.From both a macro and a micro perspective,this hybrid method produces more finely-grained insights when analyzing the dynamic dimensions of a technological field.Fourth,building a method for exploring multiple key-route main pathways by examining the patent families in a patent citation network.To some extent,the citation relationships represent the transitions from one sub-technology to another.Constructing a patent citation network based on patent families rather than individual patents improves the scale and density of the citation network to more fully discover potential citations.Further,distinguishing the citations provided by patent examiners and inventors with a weighting increases the possibility of locating further important patents relevant to the key evolutionary pathways.Moreover,the thesis explores the process of extracting multiple key-route main paths from the global perspective to accurately depict the context of a technology's development and reveal the significance of each of the different branches in a technology's evolution.