A Study On Policy Innovation And Diffusion Of Informatization Policy In China

Informatization is a development model of information technology and information resources with the society,economy and culture,which is the global wave of social change that emerged in the 1990s.As the main means of national guidance and regulation on informatization,informatization policy is on behalf of the national or local strategic needs and social norms.At present,researches on informatization policy focus on the evolution or change of the policy theme,and usually tracks the high-frequency concepts or core concepts.That means a lot of characteristic concepts or innovative concepts are not fully embedded in the analysis system of policy evolution.In addition,since the Ninth Five-Year period,a considerable number of policy texts have been accumulated,which makes quantitative analysis of the informatization policy texts possible.Therefore,based on the theory of policy diffusion and policy text analysis,this paper constructs a model of medium and low frequency concepts of informatization policy,including administrative dimension,diachronic dimension and classified topic dimension.Through policy text corpus acquisition and topic modeling,it explores the diffusion characteristics from administrative dimension,diachronic dimension and topic dimension.That is two typical research questions:the topic structure of informationization policy and its evolution and the structure of the topic network.The types of informatization policy texts in China are rich and can be divided into six categories,such as local decrees,informatization planning,implementation opinions or detailed rules for informatization,measures for informatization projects,informatization-related policies and informatization special policy.Different policy texts have different document formats,versions and structures.This paper takes governments issued strategic informatization planning documents as research object.According to the principles of pertinence and timeliness,the text collection range is determined,and 435 policy texts are obtained by hierarchical acquisition and multi-source acquisition.Then,policy text cleaning is based on the principles of pertinence integrity and availability and ultimately 412 valid samples are achieved.The topic modeling procedure involves text segmentation,part-of-speech tagging,semantic merging and low frequency words removing.Before text segmentation,the paper customizes stop words list and filtered vocabulary to improve segmentation effect.Using ICTCLAS for part-of-speech tagging in order to filter substantive words and part of the verbs.To combine synonyms and homoionyms,the semantic merging is based on the Synonymy Thesaurus.Then,remove the words that frequency is below a certain threshold.Finally,alternative optimization word set of the informatization policy has been obtained.Lastly,evaluate the validity of the topic modeling method.Based on the semantic tagging and its semantic dictionary of the Synonymy Thesaurus,the paper analyzes characteristics of diffusion about structure of policy elements on the informatization policy,including policy subject,policy object and policy behavior from administrative dimension and diachronic dimension.Also,the disadvantages of the Synonymy Thesaurus are put forword Based on the construction of the topic-optimized word set of informatization policy,the mapping mechanism of the "alternative word set-candidate word set-alternative optimization word set" is established,which can further carry out the subject index for the policy sample individually and form the vector space model of the policy samples and the characteristics of the policy text description framework.Through the characteristics of policy text-topic distribution,the similar or relevant policy topic network and the features-theme co-occurrence network are constructed,from functional and pragmatic aspects to identify the similarity of topics of informatization policy.Furthermore,the structure of overall network and cohesive subgroup and topic position are analyzed.And further discuss policy role classified and the issue of related subjects' diffusion.Also,analyze the characteristics of topic diffusion of informatization policy from administrative dimension and diachronic dimension.The shortcomings of this paper lie in word segmentation tool and policy text sample.During natural language processing of the policy text,because there is no standardized controlled vocabulary system about information policy,the common word segmentation tool must be used,that the analysis efficiency will affect the results;the second is that the completeness of the samples will influence on measuring the diffusion index.The future research prospects are:Firstly,more reliable word segment and policy text processing tool need to be provided,such as the controlled standard vocabulary,informatization policy ontology,etc.;Secondly,the informatization policy texts need to be further collected because the content analysis of all samples will be able to better understand and analyze the informatization policy hotspot and development trends;Thirdly,more samples or topics are to verify the reliability of policy diffusion analysis and put retrospective research to a more general aplication scenarios and applied to policy needs analysis,prospective policy research and comparison.