Novelty and impact: Using document similarity to study important inventions in historical Swedish patents, 1890–1929

The paper presents a method that studies similarity between patent documents to investigate what past inventions have been important. This question of assessing patent value is particularly challenging in historical contexts, where relevant metadata used for valuation such as patent citations are not available. The idea behind the method is that similarity between patent documents allows us to estimate the novelty of a patented invention (new technical vocabulary) and its impact for the subsequent inventions (vocabulary is repeated). Being both novel and impactful signals an important invention, a ”breakthrough” patent. We apply the method to analyse historical patent documents granted in Sweden between 1890 and 1929, that have been published in the Swedish Historical Patents database. The paper builds on previous work by Kelly et al. (2021), where they apply the method to study U.S. patent data. We improve the method in terms of data preparation and normalization, and use the method to examine technological advances at an aggregate level over time in Sweden. We conclude that the data preparation is an important step to be able to focus on the relevant vocabulary that regards the technological content of the patents. This measure of importance can be applied in the future to study, for example, the study of the backgrounds of the inventors with important patents.