| 俞琰,董优优,刘攀,杨舸.融入技术匹配的高校可转化专利识别研究[J].数字图书馆论坛,2026,22(4):78~88 |
| 融入技术匹配的高校可转化专利识别研究 |
| Research on Identifying University Transferable Patents Integrated with Technology Matching |
| 投稿时间：2026-03-03 |
| DOI：10.3772/j.issn.1673–2286.2026.04.008 |
| 中文关键词: 可转化专利；高校；企业；技术匹配；技术互补；技术相似 |
| 英文关键词: Transferable Patent; Universities; Enterprises; Technology Matching; Technology Complementarity; Technology Similarity |
| 基金项目:本研究得到国家社会科学基金一般项目“数据驱动的高校技术转移供需信息挖掘模式构建研究”（编号：23BTQ098）资助。 |
| 作者 | 单位 | | 俞琰 | 南京工业大学图书馆；南京工业大学经济与管理学院 | | 董优优 | 南京工业大学经济与管理学院 | | 刘攀 | 南京工业大学经济与管理学院 | | 杨舸 | 南京工业大学图书馆 |
|
| 摘要点击次数: 5 |
| 全文下载次数: 0 |
| 中文摘要: |
| 高校可转化专利识别是提升高校科技成果产业化水平的重要课题。然而，面对大量高异质、技术话语密集的专利数据，如何从中有效识别出具备转化潜质的专利，仍缺乏成熟、可操作的评价工具和筛选机制。目前相关研究多基于机器学习挖掘高校专利本身特征，侧重于评估高校专利的内在价值，忽略了高校专利与受让企业之间的技术供需适配性。为此，本研究提出一种融入技术匹配的高校可转化专利识别方法：首先构建融入技术匹配的可转化专利识别特征，其次构建应对稀疏专利转化场景的高校专利转化困难负采样，再次构建多算法优势的高校可转化专利加权集成识别模型，最后基于博弈归因分析高校可转化专利驱动因素分析。本研究选取膜技术领域开展高校可转化专利识别实验，结果表明该识别方法相较于现有仅关注内在价值的识别方法，识别准确性显著提升，展现出更好的工业实际应用价值。 |
| 英文摘要: |
| Research on the identifying of transferable university patents is an important topic for enhancing the industrialisation level of scientific and technological achievements in universities. However, faced with large volumes of highly heterogeneous and technically dense patent data, there remains a lack of mature and operable evaluation tools and screening mechanisms to effectively identify patents with transfer potential, and related work is still in the exploratory methodological stage. Current research predominantly employs machine learning to mine the intrinsic characteristics of university patents, focusing on assessing their inherent value while neglecting the technological supply-demand compatibility between university patents and potential corporate assignees. To address this gap, this study proposes a technology-matching-based approach for identifying transferable university patents. First, this study constructs identifying features integrated with technology matching. Then, designs a negative sampling method for difficult university patent transformation in sparse patent transformation scenarios. Next, constructs a weighted ensemble identifying model for university transferable patents that leverages the advantages of multiple algorithms. Finally, analyzes the driving factors of university transferable patents based on game-theoretic attribution. Experiments on transferable patent identifying in the membrane field demonstrate that the proposed method is both feasible and effective. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
