This repository relates to the working paper entitled "An evolutionary view on the emergence of Artificial Intelligence", available in http://arxiv.org/abs/2102.00233. The current repository is based on two codes, described below:
- The code "Fields_analysis" is used for analysing patents using the technological fields' classification, which is also the one used for calculating the measures of complexity and relatedness presented in the paper; this code generates Figures 1, 2, 3, 4, 5, 7, 8, and 9, presented in the paper;
- The code "4-digits_analysis" is used for analysing patents using the 4-digits IPC classification, a.k.a. subclass codes; this code generates Figure 6, presented in the paper;
Moreover, there are three folders in this repository:
- a folder named "Files_created_with_the_code", which contains all data ("Files_created_with_the_code/data") and figures ("Files_created_with_the_code/figures") generated through the codes;
- a folder named "other_files", which contains additional information needed for running the codes. This information includes a file ("ipc_technology.csv") used to translate 4 digits IPC codes into IPC technological fields (used in the code "Fields_analysis"); a file with all AI patents identifying through our keyword-based search ("IPCs_AI"), and a file containing labels used for defining the labels used in Figure 6 of the paper ("Summary IPC labels").
- a hidden folder named "large_files", which contains the large files with all priority patents used in the analysis, and is defined through a ".gitignore" file to not be updated. These big files (i.e. All_patents_and_IPCs_Part1, All_patents_and_IPCs_Part2, All_patents_and_IPC_codes_Part1, and All_patents_and_IPC_codes_Part2) are not made publicly available because they comprise a significant share of data owned by Patstat 2019. A small sample of them is available in the folder "other files". The file "All_patents_and_IPCs_Part1_SmallSampleWithHeader" contains a sample of the files used in the code "Fields_analysis", whereas the file "All_patents_and_IPC_codes_Part1_SmallSampleWithHeader" contains a sample of the files used in the code "4-digits_analysis". You can build these datasets by applying the strategy presented in De Rassenfosse et al., 2019 (https://doi.org/10.1038/s41597-019-0264-6) for identifying first filings and the geolocation of their respective inventors to Patstat 2019 spring version.