Research On Industrial Development Based On Topic Mining And Technology Path Visualization

This is an era of explosive technological growth.Emerging technologies such as Cloud Computing,Virtual Reality,Machine Vision,and Auto Pilot are emerging one after another.The emergence of disruptive emerging technologies often replaces a large number of traditional technologies.So,correctly identifying the trends of industry development will help to seize the opportunity of industrial development?Traditional industrial technology research,mainly based on expert interviews,is likely to cause subjective bias.The emergence of bibliometrics has somewhat alleviated this situation.However,the indicator-based bibliometrics often ignores the semantic information in scientific literature data(such as papers and patents).On the other hand,the exponential growth data is also a huge challenge to bibliometrics.In response to the above problems,this thesis first proposes a new topic model algorithm extended from HDP(Hierarchical Dirichlet Processes),and the proposed topic model is named CIHDP(Citation Involved Hierarchical Dirichlet Processes).In this model,the citation information is used to influence the topic modeling process of HDP,which can help mining more coherent topics for scientific literatures data.Then,based on the open domain information extraction technology,a new technology path visualization method “Topic-Problem-Method” ThemeRiver is proposed.This method can not only show the evolution trend of each technology in the process of industry development,but also show the evolution of research problems and methods in the process of technology development.Related research methods are used to identify the technological development process of the robot industry,and successfully identifying core technologies in the development of industrial robots.Moreover,by the “Topic-Problem-Method” ThemeRiver,both the trend of technology development and the evolution of their research problem and method is identified successfully.