ai_detect_for_sanqian

What problem does it solve?

Helps you reduce AIGC-detection risk in Chinese academic theses by diagnosing “AI fingerprints” and rewriting only the flagged parts without inventing data or conclusions.

Core Features & Use Cases

• Detection-and-fix workflow: Supports scanning/diagnosing (mode A), iterative multi-round rewriting (mode B), and prioritizing from ViP reports (mode C).
• Three-stage progressive rewrite: Uses a stepwise approach—wording replacement, “sentence burstiness” injection, then structural reorganization—targeted at suspicious (e.g., “purple”) paragraphs.
• Thesis-safe constraints: Enforces “no fabrication” (no new data/claims/citations) and focuses on expression changes while preserving technical meaning.
• ViP-oriented strategy: Tailors rewrites to the way ViP evaluates text features such as perplexity, burstiness, and structural patterns.

Quick Start

Ask your AI coding assistant: “降AI率，扫描我这篇论文的AI痕迹，并按三轮递进法只改标紫的段落到维普检测率低于目标值。”