tm-vocabulary · papers · online appendix · the fifty themes · IPO viewer · code

Vocabulary position in the US trademark record

Every US trademark application describes the goods or services the mark will cover, and every registered mark's owner must later swear that the product is still being sold. This project re-parses all 13.99 million USPTO trademark case files into dated events, and scores each goods/services description on two axes: atypicality, how unusual its language is for its industry, and lead, whether its language ran ahead of or behind where the industry's language was moving. Both compare the description's mix of fifty themes with its own Nice class's filings in the five years before and the five years after it was filed.

Papers

• An Event-Dated Corpus of US Trademark Prosecution and a Two-Sided Measure of Vocabulary Position (SSRN 7520758) — the corpus, the measure, and its validation.
• When Does It Pay to Be Early? Evidence from Every US Trademark Filing (in preparation) — whether being early pays: product survival, funding, listing and acquisition, and where the cost of leading is concentrated.
• Business Themes in the Trademark Record (SSRN 6954598) — the earlier combined working paper.

Data

The derived tables (dated prosecution events, event-dated five-year-proof outcomes, the theme model and per-filing scores) are archived at Zenodo: doi:10.5281/zenodo.22963531. Code is on GitHub.

Online appendix

• Per-industry results — the five-year-proof contrast and registration profiles for every Nice class
• The fifty themes — what a theme is, how they were found, and each theme's share of every class
• IPO viewer — every class's registrations on the lead/atypicality plane, colored by the furthest outcome reached
• 500-theme explorer — the finer model used as a robustness check
• Code repository (GitHub)

← aporia.institute