7toCR/paper2patent

What it solves

Researchers often struggle to convert their academic papers into legal patent applications due to the high professional barrier of patent writing, which requires specific legal language, structured claims, and strict formatting rules. This project provides a systematic way to bridge the gap between academic contribution (papers) and legal protection (patents).

How it works

The project provides a set of sophisticated LLM prompts (Flash and Pro versions) that act as a professional patent attorney. These prompts guide an AI to:

1. Extract Core Contributions: Identify the technical difference, the problem solved, and the resulting effect from a research paper.
2. Restructure Content: Convert academic descriptions into a five-part patent structure: Abstract, Abstract Drawing, Claims, Specification, and Drawings.
3. Enforce Legal Constraints: Apply strict rules for Chinese patent law, such as banning uncertain words (e.g., "about", "approximately") and enforcing specific punctuation rules for claims.
4. Generate Visuals: Includes a specialized prompt for Gemini to create compliant black-and-white line drawings and flowcharts based on the generated text.

Who it’s for

• Academic researchers in AI, Computer Vision, Deep Learning, and NLP who want to protect their inventions.
• Patent engineers and university research teams looking for a standardized template to accelerate the conversion process.

Highlights

• Legal Compliance: Strictly follows the 2020 revised version of the Chinese Patent Law and its implementation rules.
• Structured Workflow: Uses a "Difference $\rightarrow$ Problem $\rightarrow$ Effect" logic chain to ensure patentability.
• Strict Fidelity: Explicitly forbids the AI from inventing or extending technical details not present in the original paper.
• Hierarchical Protection: Guides the creation of independent and dependent claims to establish a layered legal defense.