Scholarlink

Advanced technological development across various sectors requires comprehensive analysis of global innovation landscapes and access to cutting-edge technological foundations for creating next-generation goods and services. Effective enterprise growth in today’s interconnected world necessitates strategic collaboration with partners sharing complementary technological expertise. Potential collaborators can be identified and prioritized based on the strategic value of their patented technological innovations, enabling further ranking of these partners through the significance of their intellectual property contributions. The scientific novelty lies in the developed method of analyzing a patent array as textual big data to determine the prospects of patented technologies, for the first time using the cooperation of the proposed metrics of innovation potential: the mass nature of the subject of the patented invention in the current period, the predicted mass nature of the subject (technology) in the future period, the success of a patent in the information field. The developed program provides the following functions: a) parsing patent documents from the Google Patents system, which has an extended set of parameters, and in the absence of the IPC classification, extracting it from the Yandex.Patents system; b) expanding the patent array through the analysis of IPC classes and patent citations corresponding to the areas of interest of enterprises in the Volgograd Region; c) storing patent description elements in the PostgreSQL DBMS; d) determining the mass nature of the subject of the patented invention in the current period based on the lists of keywords, by which term-frequency vectors are constructed for further clustering by topic, and reducing the dimensionality of the term-frequency matrix based on the TruncatedSVD and UMAP algorithms; e) determining the predicted mass nature of the subject in the future period for the following year based on the ARIMA model; f) determining the success of a patent in the information field based on the Google PageRank algorithm; g) extracting criteria in JSON format.

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

The New Intelligent Method of Patent Array Analysis

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Alexander A. Rublev,
  • Dmitriy M. Korobkin,
  • Alexander B. Golovanchikov

摘要

Advanced technological development across various sectors requires comprehensive analysis of global innovation landscapes and access to cutting-edge technological foundations for creating next-generation goods and services. Effective enterprise growth in today’s interconnected world necessitates strategic collaboration with partners sharing complementary technological expertise. Potential collaborators can be identified and prioritized based on the strategic value of their patented technological innovations, enabling further ranking of these partners through the significance of their intellectual property contributions. The scientific novelty lies in the developed method of analyzing a patent array as textual big data to determine the prospects of patented technologies, for the first time using the cooperation of the proposed metrics of innovation potential: the mass nature of the subject of the patented invention in the current period, the predicted mass nature of the subject (technology) in the future period, the success of a patent in the information field. The developed program provides the following functions: a) parsing patent documents from the Google Patents system, which has an extended set of parameters, and in the absence of the IPC classification, extracting it from the Yandex.Patents system; b) expanding the patent array through the analysis of IPC classes and patent citations corresponding to the areas of interest of enterprises in the Volgograd Region; c) storing patent description elements in the PostgreSQL DBMS; d) determining the mass nature of the subject of the patented invention in the current period based on the lists of keywords, by which term-frequency vectors are constructed for further clustering by topic, and reducing the dimensionality of the term-frequency matrix based on the TruncatedSVD and UMAP algorithms; e) determining the predicted mass nature of the subject in the future period for the following year based on the ARIMA model; f) determining the success of a patent in the information field based on the Google PageRank algorithm; g) extracting criteria in JSON format.