基于文本关联分析的学术论文政策影响力评价研究——以人工智能领域为例
Evaluation of Policy Impact of Academic Achievements Based on Text Correlation Analysis--An Example of the Field of Artificial Intelligence
［目的/意义］学术论文是传播和检验科学知识的主要形式，其知识结晶进入政策能够提升文件的科学性和合理性。通过评价其政策影响力，可以有效展示学术论文如何转化为政策行动，探索学术论文影响力评价的新视角。［方法/过程］文章运用Sentence-BERT、SBERTopic等文本挖掘技术，从提及影响力、主题影响力和语句影响力三个维度，构建基于语义关联的学术论文政策影响力评价指标体系。以2010-2024年人工智能领域的论文和政策文本进行实证分析，结合熵权-TOPSIS评价法对指标赋予权重，综合评价政策影响力。［结果/结论］仅有少数论文展现出较高的政策影响力，这些论文的主题分布和提及内容有一定的偏向性；在不同评价标准下，人工智能领域论文的影响力存在显著差异。未来学者和决策者应合力破解学术成果的政策转化难题，建立健全政府智库数据平台，推动学术论文更多、更快地进入政策制定流程，探索更加全面有效的评价体系。
信息传播、知识传播科学、科学研究计算技术、计算机技术
文本关联分析学术论文政策影响力人工智能
金洁琴,颜喻惟.基于文本关联分析的学术论文政策影响力评价研究——以人工智能领域为例[EB/OL].(2025-08-27)[2025-10-26].https://chinaxiv.org/abs/202508.00408.点此复制
[Purpose/Significance] Academic papers are the main form of disseminating and testing scientific knowledge, and the crystallization of knowledge into policies can improve the scientificity and rationality of documents. By evaluating its policy impact, it can effectively demonstrate how academic papers can be transformed into policy actions, and explore a new perspective on the evaluation of academic paper impact. [Method/Process]This paper uses Sentence-BERT, SBERTopic and other text mining technologies to construct an evaluation index system of policy influence of academic papers based on semantic association from three dimensions: mention influence, topic influence and sentence influence. Based on the empirical analysis of papers and policy texts in the field of artificial intelligence from 2010 to 2024, the indicators are weighted by combining the entropy weight-TOPSIS evaluation method to comprehensively evaluate the policy influence. [Results/Conclusions] Only a few papers showed high policy influence, and the topic distribution and content of these papers were biased. Under different evaluation criteria, there are significant differences in the influence of papers in the field of artificial intelligence. In the future, scholars and policymakers should work together to solve the problem of policy transformation of academic achievements, establish and improve the data platform of government think tanks, promote more and faster academic papers into the policy-making process, and explore a more comprehensive and effective evaluation system.
Text associationanalysis academicpapers policy influenceartificial intelligence
展开英文信息
评论