| In today's era,artificial intelligence is an important driving force of the new round of technological revolution and industrial change,and is considered the next "super windfall" of technological innovation.To capture the opportunities of the new round of technological revolution and industrial change,the Chinese government has followed the new trend of change and taken the development of artificial intelligence as an important strategy to enhance China's comprehensive strength,intensively introducing a series of policies related to the development of artificial intelligence industry.Text data is the most extensive carrier of industrial technology,for China's artificial intelligence industry policy text,the conventional way of reading and analysis can no longer quickly grasp the policy direction.Based on this,this paper uses LDA models for text mining of policy data to identify hot topics and evolutionary trends from a large number of policy texts,and analyse and predict the policy tendencies and development trends of the AI industry based on the textual topics,with a view to better grasping the AI policy development process in China and providing support for the orderly development of the AI industry.This paper uses the 179 AI policy texts included in the BYU Law database from2009-2022 as a data source.Firstly,by converting policies in textual form into quantitative representations through a policiometrics approach,the number of policies over time is divided into stages and the evolutionary trends in the number of policies introduced are analysed from a macro perspective.Secondly,the Gensim Lda Model function in Python is used to build a topic model,and the optimal number of topics is determined using perplexity and LDAvis visualization techniques,calculate the topic strength based on the document-topic distribution file,and perform topic identification based on the topic-word item distribution file.Finally,the evolutionary paths of policy themes at different times are constructed through time series to show 3 types of evolution of themes: growth,weakening and stability.The study found 20 hot topics for AI policy from 2009-2022,with eight topics showing an increasing trend in intensity,nine topics showing a decreasing trend in intensity and three topics showing a stable trend in intensity.China's AI industrial policy will continue to focus on the smart manufacturing industry in the future and use smart technology innovation to give rise to intelligent service applications,accelerate the construction of AI innovation pilot zones and promote the development of smart cities.Since topics such as smart key technology,enterprise innovation and talent training have reached relative saturation and are gradually entering a convergence period,we will not focus on such topics in the short term.Topics such as data sharing systems and intelligent service areas will continue to be stable for some time.The innovation of this paper is to use a text mining perspective to calculate the text of AI industrial policies,with a view to providing more objective and feasible reference suggestions for the government or enterprises.At the same time,it enriches the theoretical framework of information analysis methods driven by textual data and provides theoretical and methodological references for the subsequent deeper development of textual data mining. |
PDF Full Text Request