The Method Of Technology Foresight Based On Patent Mining And Its Application

With the continuous acceleration of technology upgrading as well as increase of difficulties,risks and costs of technology R&D,Organizations need to think about how to focus their limited resources on the most valuable R&D areas.As an effective way to solve this problem,technology foresight has been paid more and more attention recently.As Brock-hoff^[1]pointed out,patent literature contains 90%to 95%of the latest scientific research results around the world,and these standardized data related to innovation intention and technological development can be freely used,which can be an effective tool to measure the focus and direction of innovation.Taking patent big data as mining object,this paper proposes a more effective technology foresight method based on patent mining,which combines the application domain information of IPC classification system and patent text structure and function information to build patent similarity measurement model,on the micro level of patent text structure and functional information representation method,a patent text structure and functional representation model based on SAO chain is proposed to completely present the structure and functional information of a patent;Then,patent similarity is calculated based on constructed patent similarity measurement model,and patent network as well as patent map are drawn based on obtained patent similarity matrix,technology evolution path is analyzed by studying the patent network and its future trends is predicted combining with TRIZ theory,and potential technology opportunities are identified by reading technical vacancies in the patent map;Finally,all the analysis results are combined to carry out technology foresight results refinement.The structural and functional representation model based on SAO chain proposed in this paper can completely present the structure and function information of patent text,which makes up for the deficiency of existing patent mining methods which only focus on description language but ignore the complete structure and function information of a patent,in addition,the patent similarity measurement model proposed in this paper comprehensively considers macro application field as well as micro text structure function information,which can obtain more effective patent similarity measurement results and ensure the effectiveness and reliability of subsequent analysis.Finally,the proposed method is applied to the field of industrial robot to carry out empirical research,and the technology evolution path,future trends and potential opportunities are analyzed and summarized,digging out control,programming,sensing,communication,mobile technology,interactive technology,morphological structure and operation mode of eight direction of development path and future trends,identifying the communion industrial robot system,software customization and generalization,intellisense and intelligent learning,cloud service industrial robot and modular,reconfigurable and miniaturized hardware of five potential technology opportunities,combined with the predicted results,the corresponding countermeasures and suggestions for the future development of China’s industrial robot are put forward.