The Language of Innovation

Predicting innovation is a peculiar problem in data science. Following its definition, an innovation is always a never-seen-before event, making the usual approach of learning patterns from the past a useless exercise. Here we propose a strategy to address the problem in the context of innovative patents, by defining innovation as never-seen-before associations of technologies. We think of technological codes present in patents as a vocabulary and the whole technological corpus as written in a specific, evolving language. We leverage such structure with techniques borrowed from Natural Language Processing by embedding technologies in a high dimensional euclidean space where relative positions are representative of learned semantics. Dynamics on this space predicts specific innovation events, that are tested against null models. These methods provide a completely new way of understanding and forecasting innovation, by tackling it from a revealing perspective and opening interesting scenarios for a number of applications and further analytical approaches.