Abstract: In the process of innovation, the key to the success of innovation is to identify the matched partners. Partner identification from the perspective of technology convergence can organically combine the technology matching characteristics of partners with organizational innovation strategy, improve the success rate of cooperation and improve innovation performance. Link prediction in multi-layer networks can fully mine the link information in multi-layer networks, so as to identify potential cooperative links more accurately. By constructing the multi-layer network model of technology convergence-organization cooperation, the research considers the future technology convergence trend in the process of identifying partners, and proposes the potential characteristics of technology convergence among organizations based on the technology convergence trend, and realizes the prediction of partner links in the multi-layer network based on the structural indicators, partner diversity and technology convergence potential. Through the empirical study of patents in the field of artificial intelligence from 2005 to 2020, it is verified that the prediction accuracy of partner links in the multi-layer network is significantly higher than that in the single-layer network. The research results can provide scientific reference for enterprises to choose partners when they implement different innovation strategies.

摘要： 创新过程中组织识别匹配的合作伙伴是创新成功的关键。从技术融合的角度进行合作伙伴识别能够将伙伴的技术匹配特征与组织创新战略有机结合，提高合作成功率，提升创新绩效。多层网络中的链路预测能够充分挖掘多层网络中的链接信息，从而更准确地识别出潜在的合作链接。研究通过构建技术融合-组织合作多层网络模型，在识别伙伴的过程中考虑了未来的技术融合趋势，并基于技术融合趋势提出了组织间的技术融合潜力特征，实现了多层网络中基于结构指标、伙伴多样性与技术融合潜力的合作伙伴链路预测。通过2005-2020年人工智能领域专利的实证研究，验证了多层网络中的合作伙伴链路预测精度显著高于单层网络中的预测精度。研究结果能够为企业实现不同创新战略时选择合作伙伴提供科学参考。