patccat - PATent Claim Classification by Algorithmic Text-Analysis

  • Find the data on Zenodo DOI: 10.5281/zenodo.22401669
  • Our current release is 4.0.0 (September 15, 2026).
  • The code is available as an R package (patccatClaims Latest release - September 10, 2026) on Github . 

    # install.packages("remotes")
remotes::install_github("bganglmair/patccatClaims")
    You can find more information on patccatClaims on the package page.
  • For a detailed description of our data construction and a discussion of the data and results see our paper ("The Rise of Process Claims: Evidence from a Century of U.S. Patents") on SSRN. You can find the main findings in this short handout. For a short summary of the project, take a look at this deck of slides (December 2021).
  • For more patccat-related data products, visit the project's Zenodo community page at https://zenodo.org/communities/patccat/.

patccat - EPO edition


U.S. Historic Patent Claims, 1836 - 1975 (Full Text)

  • Find the data on Zenodo. Version 1.1.0 posted on September 6, 2026.
  • This dataset presents claim-level full text for U.S. utility patents granted between 1836 and 1975. It contains 21,782,888 claims from 3,663,243 patents. It is the outcome of a multi-step process: finding the claim boundaries in historic patents, splitting the claims, correcting OCR damage with a closed correction table, and publishing one row per claim. The series stops at 1975 because patents granted from 1976 onward carry their claims as a separate, machine-readable field and need no extraction.

Mannheim Privacy Policy Panel (MaPPP)

The Mannheim Privacy Policy Panel (MaPPP) contains the texts of 1,775,388 million privacy policies posted by 311,539 German firms between 2012 and 2021. The files were collected from the Internet Archive via the Wayback Machine. The MaPPP is an unbalanced quarterly panel.

Loan Agreement Text Data: Contract Texts, Parsed Sections, and Clause Breakpoints from SEC EDGAR Filings (1996–2017)

This dataset contains the full text of private loan agreements filed with the U.S. Securities and Exchange Commission (SEC) via the EDGAR database, covering the period 1996-2017. The data accompanies: "Complexity, Standardization, and the Design of Loan Agreements" (with Malcolm Wardlaw)

Consumption-to-Industry Concordance Table

  • In "Markups for Consumers" (Ganglmair, Kann, and Tsanko), we construct a concordance table between 4-digit COICOP (consumption) classes and 4-digit NACE (industry) classes. Download the file here.
  • If you decide to use the table, please cite the paper:

    Ganglmair, Bernhard, Alexander Kann, and Ilona Tsanko (2021): "Markups for Consumers," Journal of Economics and Statistics, 241(5-6), 2021 (November), pp 701-734

  • We are collecting research output that uses our table. If you'd like to be listed, please send us the paper or citation information.
  • If you have any suggestions for improvement or if you find any errors in the table, please reach out.

Replication Data

  • "Learning When to Quit: An Empirical Model of Experimentation in Standards Development" | Published in American Economic Journal: Microeconomics. Data available through Zenodo
  • "Representation is Not Sufficient for Selecting Gender Diversity" | Published in Research Policy. Data coming soon.
  • "Markups for Consumers" | Published in the Journal of Economics and Statistics. Data and replication files available here

If you want to see more data useful for innovation research, check out the I3 Open Innovation Dataset Index.

Last changed: September 15, 2026