Scholarlink

<p>Accurately identifying high-value patents can be difficult with the dramatic increase in the number of patents applications. Traditional approaches for identifying high value patents are difficult to identify newly granted patents. The identification accuracy is low due to the imbalanced distribution of patent data. In order to solve above problems, a high-value patent identification framework that combines hybrid sampling technology and ensemble learning algorithm is proposed. In this framework, traditional high value patent identification indicator system is reconstructed by adding technical capability of patentees. In order to solve imbalanced data distribution, a combined model that integrates oversampling techniques and ensemble learning algorithms is proposed, where the ADASYN-LOF method primarily addresses class imbalance and reduces noise interference in synthetic samples. Furthermore, Genetic Algorithm (GA) are used to optimize the parameters of AdaBoost. To test the effectiveness of above model, patent data in field of scientific instruments are used. Tenfold cross-validation is carried out to evaluate the performance between our model and other 4 existing models. The results show that ADASYN-LOF-GA-AdaBoost model performs better than other models. Therefore, this model can effectively identify high-value patents with transfer potential.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Research on high-value patent identification model from perspective of patent transfer

  • 期刊论文

  • 发表日期：

    2026年3月18日

查看全文
  • Zengyuan Wu,
  • Ying Li,
  • Xiangli Han,
  • Yuanyang Teng,
  • Xueyan Niu,
  • Bin He

摘要

Accurately identifying high-value patents can be difficult with the dramatic increase in the number of patents applications. Traditional approaches for identifying high value patents are difficult to identify newly granted patents. The identification accuracy is low due to the imbalanced distribution of patent data. In order to solve above problems, a high-value patent identification framework that combines hybrid sampling technology and ensemble learning algorithm is proposed. In this framework, traditional high value patent identification indicator system is reconstructed by adding technical capability of patentees. In order to solve imbalanced data distribution, a combined model that integrates oversampling techniques and ensemble learning algorithms is proposed, where the ADASYN-LOF method primarily addresses class imbalance and reduces noise interference in synthetic samples. Furthermore, Genetic Algorithm (GA) are used to optimize the parameters of AdaBoost. To test the effectiveness of above model, patent data in field of scientific instruments are used. Tenfold cross-validation is carried out to evaluate the performance between our model and other 4 existing models. The results show that ADASYN-LOF-GA-AdaBoost model performs better than other models. Therefore, this model can effectively identify high-value patents with transfer potential.