现代情报2026，Vol.46Issue(7)：155-168,14.DOI:10.3969/j.issn.1008-0821.2026.07.012
基于概念组合特征的学术论文创新性测度研究
Measuring Innovation in Academic Papers via Concept Combination Characteristics
摘要
Abstract
[Purpose/Significance]Constructing a scientific and effective method for evaluating the innovativeness of academic papers is of great significance for improving the quality of research results and the efficiency of scientific research resource allocation.[Method/Process]Based on knowledge composition theory,this study constructed an innovativeness measurement method that includes two core dimensions:novelty and heterogeneity,comprehensively considering the co-occurrence frequency,positional importance,and semantic similarity of concept combinations.Empirical analysis was con-ducted using academic paper data from the field of knowledge management.In terms of methodological implementation,this study employed large language models to extract the core concepts of academic papers and applied strategic coordinate analysis to identify concept combinations that exhibit both high novelty and high heterogeneity as innovative concept combi-nations.The proportion of innovative concept combinations among all concept combinations within a paper is then used as the innovativeness score of the paper.[Result/Conclusion]The results show that highly innovative papers generally possess characteristics such as thematic diversity,technological novelty,and interdisciplinary integration.Further comparative verification results indicate that the method based on concept combination characteristics has a better identification effect on highly innovative papers and a more significant ability to distinguish innovative levels.关键词
创新性评价/学术论文/知识组合理论/概念组合/战略坐标分析法Key words
innovativeness evaluation/academic papers/knowledge recombination theory/concept combination/strategic coordinate analysis分类
社会科学引用本文复制引用
黄颖,杜艺璐,毛进..基于概念组合特征的学术论文创新性测度研究[J].现代情报,2026,46(7):155-168,14.基金项目
国家自然科学基金面上项目"多源数据融合视角下技术会聚的形成机制与预测评估研究"(项目编号:72374162) （项目编号:72374162）
国家自然科学基金面上项目"基于'问题—方法'关联识别的科学知识创新探测与协同演化分析"(项目编号:72174154). （项目编号:72174154）
