AutoPatent: A Multi-Agent Framework for Automatic Patent Generation

Qiyao Wang, Shiwen Ni, Huaren Liu, Shule Lu, Guhong Chen, Xi Feng, Chi Wei, Qiang Qu, Hamid Alinejad-Rokny, Yuan Lin, Min Yang · 2024

Read AutoPatent: A Multi-Agent Framework for Automatic Patent Generation by Qiyao Wang, Shiwen Ni, Huaren Liu, Shule Lu, Guhong Chen, Xi Feng, Chi Wei, Qiang Qu, Hamid Alinejad-Rokny, Yuan Lin, Min Yang (2024) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2412-09796.txt.

Description

Qiyao Wang and co-authors present a 2024 system that drafts full patent files from rough drafts with a team of AI agents. Patents run long, near seventeen thousand tokens on average, with stiff terms and set forms that strain language models. Past AI patent work stuck to sorting or short texts. The new task, Draft2Patent, asks for the whole file, and a test set scores the tries.

A planner agent maps the job, writer agents draft the parts, and an examiner agent checks the whole. Two aids help: a tree of the patent's plan and a smart lookup of past text. Patents built this way by a small open model beat those from far larger models on the tests. Agent teams may tame other long legal drafts.

Catalog record

CREATORQiyao Wang, Shiwen Ni, Huaren Liu, Shule Lu, Guhong Chen, Xi Feng, Chi Wei, Qiang Qu, Hamid Alinejad-Rokny, Yuan Lin, Min Yang
YEAR2024
COLLECTIONPAPERS
FORMATTXT
FILE SIZE51 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2412-09796

Browse the full library · About the Tower of Babel