摘要
运用融合资源传导、协同过滤以及链路预测理论对“企业-专利”关系进行系统而新颖的逻辑表达,继而精准有效地识别企业潜在的合作关系,有助于企业丰富技术维度并推动产业创新化进程。基于竞争情报思想,以钙钛矿太阳能电池领域7415个技术专利为数据基础,构建以专利资源传导为基础的动态转移推荐,以链路预测方法为基础的异质网络推荐,以专利文本为基础的协同过滤推荐。通过引入可调参数,将三者融合成推荐系统,使得企业能够结合现实场景选择需要关注的专利资源和合作伙伴。研究表明,混合推荐系统的融合视角为颠覆性技术企业潜在关系预测提供了崭新的研究方法,所构建的推荐系统实现了推荐结果“准确化+新颖化+全面化”的多方位表达,为企业未来合作提供了决策和参考。
Integrating resource transmission,collaborative filtering,and link prediction theories to systematically and innovatively express the“enterprise patent”relationship,and then accurately and effectively identify potential cooperative relationships between enterprises,helps enterprises enrich their technological dimensions and promote the process of industrial innovation.Based on competitive intelligence thinking and data from 7415 technical patents in the field of perovskite solar cells,a dynamic transfer recommendation based on patent resource transmission,a heterogeneous network recommendation based on link prediction method,and a collaborative filtering recommendation based on patent text are constructed.By introducing adjustable parameters,the three are integrated into a recommendation system,enable enterprises to select patent resources and partners that require attention based on real-life scenarios.Research shows that the fusion perspective of hybrid recommendation systems provides a new research method for predicting potential relationships in disruptive technology enterprises.The constructed recommendation system achieves multi-dimensional expression of recommendation results that are“accurate,novel,and comprehensive”,providing decision-making and reference for future cooperation between enterprises.
作者
窦路遥
周志刚
申婧
黄运聪
DOU Luyao;ZHOU Zhigang;SHEN Jing;HUANG Yuncong(School of Information,Shanxi University of Finance and Economics,Taiyuan 030006,China;Anhui Branch,Agricultural Bank of China,Hefei 230000,China)
出处
《竞争情报》
2024年第2期12-23,共12页
Competitive Intelligence
基金
山西省哲学社会科学规划课题“山西省中小企业间数据融合安全共享机制研究”(编号:2022YY097)
中国教育技术协会重点项目“财经专业背景下大数据核心技术产学研用多维一体的创新性研究”(编号:XJJ202205014)
中国高等教育学会专项课题“大数据技术与应用产教融合实践与创新”(编号:21CJZD06)的研究成果。
关键词
混合推荐
关系预测
专利文本
钙钛矿太阳能电池
新颖化
mixed recommendation
relationship prediction
patent text
perovskite solar cells
novelization