| 俞琰,马昕远,刘攀.基于专利主体特征的专利权维持期预测研究[J].数字图书馆论坛,2024,20(10):53~62 |
| 基于专利主体特征的专利权维持期预测研究 |
| Patent Maintenance Period Prediction Based on Patent Subject Features |
| 投稿时间：2024-06-06 |
| DOI：10.3772/j.issn.1673-2286.2024.10.006 |
| 中文关键词: 专利权维持期；预测；专利主体特征；集成学习；可解释性 |
| 英文关键词: Patent Maintenance Period; Prediction; Feature of Patent Subject; Ensemble Learning; Interpretability |
| 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 | | 马昕远 | 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 |
|
| 摘要点击次数: 33 |
| 全文下载次数: 114 |
| 中文摘要: |
| 针对目前专利权维持期在专利价值预测方面存在的滞后性和预测特征忽略专利主体信息的问题，提出基于专利主体特征的专利权维持期预测方法。使用专利数据集，构建包括专利发明人、专利权人、专利代理的专利主体特征，并利用基于相关性的集成学习模型预测专利权维持期，使用SHAP模型对获得的预测模型加以解释，以增强理解。通过风能转化领域专利数据的实证研究验证所提方法的可行性与有效性，模型的评估指标平均绝对误差、均方误差、决定系数分别达到0.469 2、0.933 1、0.936 8。相较于已有方法取得更为理想的预测结果，表明专利主体特征能够有效地预测专利权维持期，提高预测准确性。 |
| 英文摘要: |
| This paper proposes a patent maintenance period prediction method based on patent subject characteristics to address the current issues of feature lag and neglect of patent subject information in predicting features. The proposed method uses a patent dataset to construct patent subject features including patent inventors, patent owners, and agencies, and uses a correlation-based ensemble learning model to predict the patent maintenance period. Finally, the SHAP model is used to interpret the obtained prediction model to enhance understanding. Empirical research based on patent data in the field of wind energy conversion demonstrates the feasibility and effectiveness of the proposed method in this paper. The model achieves evaluation metrics with mean absolute error of 0.469 2, mean squared error of 0.933 1, and R2 of 0.936 8. Compared to existing methods, the model achieves more ideal predictive results, demonstrating that the features of the patent subject can effectively predict the maintenance period of patent rights, thereby enhancing the accuracy of the predictions. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
