Does prior knowledge affect patent technology diffusion? A semantic-based patent citation contribution analysis

Prior knowledge is central to the development of technology and innovation, and it is essential to understand its impact on technology diffusion. This study explores the impact of different categories of prior knowledge on patent technology diffusion using a computable and explainable patent citation contribution analysis method, which considers not only the semantic information of citation content but also the diversity of citing motivations. We categorize knowledge into concept description, theoretical derivation, method design, experimental operation and production application through the BERT model. Moreover, the contribution of citations is quantified by computing the semantic similarity between sentences using Sentence-BERT model and cos-sim. The impact of prior knowledge on patent technology diffusion is explored using zero-inflated negative binomial regression and Weibull PH regression method. Specifically, the diffusion power, diffusion breadth, and diffusion speed were calculated to quantify the diffusion. Our empirical work in carbon nanotube field suggests that (a) there is a detailed relationship between science and technology that varies across periods of technology development; (b) patent technology diffusion is related to the categories of prior knowledge, with the latter having different effects on the former due to the characteristics of prior knowledge and technology development. These findings and the method could contribute to understanding and prediction of patent technology diffusion and technology development from a new analytical perspective.