Text Mining To Measure Novelty And Diffusion Of Technological Innovation

Existing measures of patent novelty and diffusion mainly rely on patent classification or citations. Given that inventive ideas are embedded in the texts of patents, our study provides a new way to assess novelty and diffusion by text mining techniques. As a validation test, we collect a set of patents linked to famous awards such as the Nobel prize. Overall, text-based measures outperform other commonly-used novelty and diffusion metrics.