Scholarlink

<p>This paper proposes a new method for matching patents with academic publications to create patent-paper pairs (PPP). These pairs can identify instances where a research result is both applied in a patent and published in a paper. The study focuses on an exhaustive sample of patenting universities and public research institutes in Japan, utilizing a new dataset that contains patent-to-article citations and a machine learning model as part of the matching process. Expert consultations and benchmarking with other models were conducted to enhance the robustness of the methodology. Using a set of 115 Japanese universities and 22 public research institutes together with patent (USPTO) and publication data (OpenAlex) between 2004 and 2018, we built a dataset of 16,899 PPPs out of 10,896 granted patents and 652,610 publications. The results demonstrate that this phenomenon is widespread in academia and our data show the diversity of the academic disciplines and technical field involved, highlighting the intricate connections between scientific and technical concepts and communities. On the methodological side, we documented in-depth complementary validation techniques to enhance the precision and reliability of our matching algorithm. Using open-source data, our methodology is adaptable to diverse national contexts and can be readily adopted by other research teams investigating similar topics.</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

‘‘Exploring academic patent-paper pairs: a new methodology for analyzing Japan’s research landscape’’

  • 期刊论文

  • 发表日期：

    2025年3月9日

查看全文
  • Van Thien Nguyen,
  • Rene Carraz

摘要

This paper proposes a new method for matching patents with academic publications to create patent-paper pairs (PPP). These pairs can identify instances where a research result is both applied in a patent and published in a paper. The study focuses on an exhaustive sample of patenting universities and public research institutes in Japan, utilizing a new dataset that contains patent-to-article citations and a machine learning model as part of the matching process. Expert consultations and benchmarking with other models were conducted to enhance the robustness of the methodology. Using a set of 115 Japanese universities and 22 public research institutes together with patent (USPTO) and publication data (OpenAlex) between 2004 and 2018, we built a dataset of 16,899 PPPs out of 10,896 granted patents and 652,610 publications. The results demonstrate that this phenomenon is widespread in academia and our data show the diversity of the academic disciplines and technical field involved, highlighting the intricate connections between scientific and technical concepts and communities. On the methodological side, we documented in-depth complementary validation techniques to enhance the precision and reliability of our matching algorithm. Using open-source data, our methodology is adaptable to diverse national contexts and can be readily adopted by other research teams investigating similar topics.