Scholarlink

<p>This study introduces a framework for constructing a dynamic and predictive innovation collaboration knowledge graph that enables enterprises to better understand the innovation landscape and identify potential collaborators. By integrating multilayer patent collaboration relationships—joint applications, transfers, and licenses—the framework offers a holistic view of innovation collaboration, transforming the static and illustrative networks into dynamic and predictive ones. The process begins with the development of an International Patent Classification co-occurrence matrix, followed by principal component analysis to identify enterprise technology clusters. A knowledge graph is then constructed using patent data from these clusters, and entities are aligned to improve consistency. The framework identifies core entities through relationship strength and control strength and predicts collaborators based on path length and path strength. Empirical results show that this framework outperforms single-behavior networks by incorporating multilayer collaboration relationships from the enterprise’s perspective, enhancing path strength by 428.06% and improving collaborator recommendations by 57.14%. Theoretically, this study contributes to the development of innovation collaboration network theory by offering a dynamic and comprehensive perspective. It integrates multilayer collaborations to facilitate predictive capabilities and examines the variations between different single behavior networks, confirming their varying orientations towards long-term and short-term collaborations. Practically, the study offers a data-driven and automatic tool for technology managers to identify potential collaborators effectively. This shifts away from traditional reliance on subjective judgments, helping to reduce bias in identification conclusions due to incomplete knowledge. By integrating diverse collaboration patterns into a unified framework, the study ultimately enhances enterprise collaboration strategies and fosters predictive advantages in today’s competitive innovation environment.</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

A novel method to identify potential collaborators with a patent-based knowledge graph

  • 期刊论文

  • 发表日期：

    2025年8月30日

查看全文
  • Jingzhu Wei,
  • Tongrui Zhang

摘要

This study introduces a framework for constructing a dynamic and predictive innovation collaboration knowledge graph that enables enterprises to better understand the innovation landscape and identify potential collaborators. By integrating multilayer patent collaboration relationships—joint applications, transfers, and licenses—the framework offers a holistic view of innovation collaboration, transforming the static and illustrative networks into dynamic and predictive ones. The process begins with the development of an International Patent Classification co-occurrence matrix, followed by principal component analysis to identify enterprise technology clusters. A knowledge graph is then constructed using patent data from these clusters, and entities are aligned to improve consistency. The framework identifies core entities through relationship strength and control strength and predicts collaborators based on path length and path strength. Empirical results show that this framework outperforms single-behavior networks by incorporating multilayer collaboration relationships from the enterprise’s perspective, enhancing path strength by 428.06% and improving collaborator recommendations by 57.14%. Theoretically, this study contributes to the development of innovation collaboration network theory by offering a dynamic and comprehensive perspective. It integrates multilayer collaborations to facilitate predictive capabilities and examines the variations between different single behavior networks, confirming their varying orientations towards long-term and short-term collaborations. Practically, the study offers a data-driven and automatic tool for technology managers to identify potential collaborators effectively. This shifts away from traditional reliance on subjective judgments, helping to reduce bias in identification conclusions due to incomplete knowledge. By integrating diverse collaboration patterns into a unified framework, the study ultimately enhances enterprise collaboration strategies and fosters predictive advantages in today’s competitive innovation environment.