Avoiding the pitfalls of direct linkage: A novelty-driven approach to measuring scientific impact on patents

Abstract

Scientific knowledge plays a major role in the generation of new technological knowledge. We present a new novelty-driven approach to measure the influence of science on patents. We overcome the weaknesses of previous methods based on either citations or semantic similarities, both representing direct linkages between documents. We combine patent novelty measurement with technology-specific, scientific dictionaries, which allow us to measure a patent's nearness to science by stable indirect linkages. We apply our indicator “science-driven novelty” to the testbed of RFID technology and confirm its validity by conducting an expert survey. Subsequently, we test how science impacts patent value, finding that scientific influence increases the average value of a patent. Our results suggest several implications. For academics, we recommend not relying solely on analyzing direct links between papers and patents to determine the influence of science on technology. For management, we provide a new tool to assess scientific influences in patents and thus the value of their company's own patent portfolio as well as the portfolios of third parties. Using text as data, the tool is viable at a very early stage and can be helpful in go/no-go decisions for technology management.