摘要
[目的/意义]提出自动化的政策文本量化方法,探索政策对科研选题的影响,为趋势预测、前沿识别以及未来的科研选题提供参考。[方法/过程]文章以战略性新兴产业中的新能源汽车领域为样板,综合使用了文献调研法、专家咨询法与政策工具,将政策文本表征为词向量后,通过特征扩展来识别关键性政策文本,并抽取政策关键短语来代表政策主题;选用LDA模型提取论文主题来表征科研选题,通过对比分析政策文本主题与论文主题来探索政策对科研选题的影响。[结果/结论]通过实证研究表明,文章提出的方法可以实现政策文本的自动分解;证实了政策会对科研选题产生一定程度的影响,并为前沿识别与趋势预测研究提供建议。
[Purpose/Significance]This article attempted to propose an automated policy text quantification method,explored the impact of policies on scientific research topics,and provide references for trend forecasting,frontier identification,and future scientific research topics.[Method/Process]The paper took the new-energy vehicle field in the emerging sectors of strategic importance as an example,comprehensively used the literature research method,expert consultation method,and policy tools,characterized the policy text as a word vector,and then used feature expansion to identify key policy texts,extracted the policy keyphrases represent the theme of the policy.We used the thesis theme,choose the LDA model to extract the topic,through the comparative analysis of the public policy and thesis topic to explore the influence of policies on scientific research topics.[Results/Conclusions]The empirical research shows that the method proposed in the article can realize the automatic decomposition of the policy;it proves that the policy has a certain degree of influence on the selection of scientific research topics,and provides suggestions for frontier identification and trend prediction research.
作者
梁继文
杨建林
王伟
Liang Jiwen;Yang Jianlin;Wang Wei(School of Information Management,Nanjing University,Nanjing 210023,China;Jiangsu Key Laboratory of Data Engineering & Knowledge Service,Nanjing 210023,China)
出处
《现代情报》
CSSCI
2021年第8期109-118,共10页
Journal of Modern Information
关键词
政策文本量化
政策分解
主题分析
科研选题
LDA模型
情报服务
前沿识别
趋势预测
新能源汽车
quantitative analysis of policy
policy decomposition
topic analysis
scientific research topic
LDA model
intelligent service
frontier identification
trend prediction
new energy vehicle
