摘要
[目的/意义]现行基于"数量"与基于"引文"的评价方法本质上是对学术论文"影响力"的评价,与学术创新性并不完全对等,从语义层次及语言结构的角度对学术论文进行创新性测度为构建新型学术论文创新性指标及方法体系提供了新思路。[方法/过程]利用自然语言处理方法,改进传统的SAO结构模型,提出了一种基于SAO-ADV模型及语义相似度的论文创新性测度方法,并以"Scientometrics"期刊2019年发表的论文为样本进行实证检验。[结果/结论]实验结果表明,排名靠前的论文多为理论上的创新,排名中间的论文则多是方法上的改进或是研究对象的创新,而Review类型的论文的排名相对靠后。同时发现,基于SAO-ADV模型及语义相似度的论文创新性测度方法测算的创新度与论文主题的创新度存在正相关关系,对客观有效地进行论文创新性评价具有参考价值。[局限]该方法没有对研究领域、研究对象、研究方法、研究理论等不同方面的创新赋予不同的权值,但是在实际的研究中,不同方面的创新是存在着差别的。
[Purpose/significance]The current evaluation method based on"quantity"and"citation"is essentially an evaluation of the"influence"of academic papers,which is not completely equivalent to academic innovation.The innovation measurement of academic papers from the semantic level and language structure provides a new idea for constructing innovative indicators and methodologies for new academic papers.[Method/process]Using the natural language processing method to improve the traditional SAO structure model,an innovation measurement based on SAO-ADV model and semantic similarity was proposed,and conducting an empirical test on the papers published in the journal"Scientometrics"in 2019.[Result/conclusion]The experimental results show that the papers ranked top are mostly theoretical innovations,while the middle papers are mostly method improvements or research object innovations,while the review papers are relatively lower.Based on the SAO-ADV model and semantic similarity,there is a positive correlation between the innovation degree calculated by the method of measuring the innovation degree of the paper and the innovation degree of the topic of the paper,which has a reference value for the objective and effective evaluation of the innovation of the paper.[Limitations]This method does not give different weights to the innovation of research field,research object,research method,research theory and other aspects,but in the actual research,there are differences in the innovation of different aspects.
出处
《情报理论与实践》
CSSCI
北大核心
2020年第11期129-136,共8页
Information Studies:Theory & Application
基金
国家社会科学基金一般项目“基于科学的创新驱动新兴技术生成的机理及其促进政策研究”(项目编号:17BGL031)
广东省科技计划软科学项目—重点项目“基于科学大数据的新能源领域前沿技术预测研究”(项目编号:2019B101001024)的成果之一。
