Inventors' Explorations Across Technology Domains

Data accompanying the paper "Inventors' Explorations Across Technology Domains". 

Code for this study is at https://doi.org/10.5281/zenodo.1035448 and https://github.com/jeffalstott/inventorexploration and as Supporting Information at the Design Science website. While that is the code and quite slim (~14MB), the data here is large (14GB zipped, ~49GB unzipped). To use, download the code from any of those sources; it will have a file structure that includes a folder data/. Then download the data here and unzip it into data/. Now you have everything you need to reproduce the study! Below is the abstract of the study:

"Technologies are created through the collective efforts of individual inventors. Understanding inventors' behaviors may thus enable predicting invention, guiding design efforts or improving technology policy. We examined data from 2.8 million inventors' 3.9 million patents and found most patents are created by ``explorers'': inventors who move between different technology domains during their careers. We mapped the space of latent relatedness between technology domains and found explorers were 250 times more likely to enter technology domains that were highly related to the domains of their previous patents, compared to an unrelated domain. The great regularity of inventors' behavior enabled accurate prediction of individual inventors' future movements: a model trained on just 5 years of data predicted inventors' explorations 30 years later with a log-loss below 0.01. Inventors entering their most related domains was associated with patenting up to 40\% more in the new domain, but with reduced citations per patent. These findings may be instructive for inventors exploring design directions, and useful for organizations or governments in forecasting or directing technological change."