基于二分网络模型的专利权人推荐研究--以新能源汽车领域为例?

通过推荐技术为企业推荐合作者，在提高技术研发效率方面有着重要的意义。在5种(物质扩散算法、热传导算法、偏热传导算法、混合算法、接受者能力算法)基于二分网络推荐算法的基础上，本文提出二阶同向资源扩散算法；同时使用β来判断对合作者合作倾向。以新能源汽车领域专利权人推荐为例进行实验，实验结果证明使用二阶资源扩散算法比一阶资源扩散算法推荐效果要好，准确率最高可达27.59%，召回率最高可达30.05%，提升幅度最大可以达到15.17%，最优β表明优先选择其曾经有过合作关系的专利权人进行合作。

The Recommendation of Patentee Based on Bipartite Network--A Case Study of New Energy Vehicles

To recommend suitable collaborators through recommendation has important significance of improving the efficiency and the quality of technology research and development. Based on five different bipartite network recommendation algorithms(diffusion algorithm, heat conduction algorithm, partial heatconduction algorithm, hybrid algorithm, the receiver ability algorithm), this paper proposed a new algorithm hereafter, we call it “the second order resource diffusion algorithm of the same direction” , and this study employed a parameter β to evaluate the tendency of cooperation. We applied this new algorithm in the ifeld of new energy vehicles as an example. The experiment results showed that the performances of this new algorithm are better than the ifrst order resource diffusion algorithm, the precision rate is 27.59%, the recall rate is 30.05%, the highest improvement is 15.17% compared with other algorithms. In addition, the results of optimal βindicated that the patentee are preference of former collaborators.