Improved multi-criteria decision making method integrating machine learning for patent competitive potential Evaluation：A case study in water pollution abatement technology

Abstract

Water Pollution Abatement (WPA) plays a pivotal part in ensuring water safety, but the key to developing WPA technology is a small number of high-value patents. This paper proposes an improved multi-criteria decision-making (MCDM) method integrating machine learning, which is applied to patent evaluation for the first time to effectively distinguish between patents of different categories of competitive potential, aiming to deepen the value understanding of the WPA industry technology. In this method, the set of reference points in AHPSort II is improved by integrating machine learning from the equidistant to automatic searching, and the original linear interpolation is modified by the curvilinear interpolation of cubic polynomial, which is more in line with the characteristics of patent data and improves the accuracy of the estimates. As an example, the 135 patents granted by the USA in WPA technology are presented: 5 patents at the T1 level, 87 patents at the T2 level, and the remaining 43 T3 level patents were identified by this method. Finally, through analyzing these T1 high-value patents, the inspirations for the development of WPA technology in China are put forward.