Using the node importance of patent network to evaluate patent relational value

Abstract

Citation serves as a common and considerable metric for evaluating the relational value between patents and technologies. This relational value, generally, can be measured by the centrality of the patent citation network. Some centrality indicators can indicate the importance of patents in the citation network, but they ignore the structural information of neighborhood patents. The structural importance of patent network is defined and calculated by considering the degree of similarity between patents and their neighboring node pairs. Briefly, we pair the “neighbor patent” of the target patent and the “neighbor patent” of these “neighbor patents”, called “node pair”. On this basis, we measure the relational value of the target patents. The structure analysis method of patent citation network improves patent value evaluation method from a network science perspective. Firstly, a comprehensive patent citation network is constructed. Secondly, the degree of similarity of patents and their node pairs is used to characterize their local network structural importance, and based on this, PNII, a patent node importance index, is proposed for patent value evaluation. Finally, we applied SIR model to calculate the actual propagation influence of patents, which is used as a criterion to compare the evaluation effect of PNII and other centralities. The patent relational value evaluation result shows that the PNII based on the node importance of patent network is more scientific and accurate than the general network centralities.