JoyEdit Academic AI Trace Removal

Problem It Solves

Chinese academic text rewritten by LLMs often leaves three kinds of traces: inflated significance, mechanically uniform sentence rhythms, and chatbot-style formatting residue. Phrases such as “In the era of...”, “not only... but also...”, “quality and efficiency”, “double-edged sword”, and lists like “First, theoretical basis: ...” can make a paper look like generic model output rather than peer-reviewed scholarly writing. JoyEdit Academic AI Trace Removal treats these issues as rewriting constraints instead of performing simple synonym substitution.

How It Works

The skill calibrates toward CSSCI/core-journal style and runs an auditable rewrite process:
- Detect AI patterns: scans 33 Chinese LLM writing traits, including grand narratives, vague citations, parallelism stacking, promotional wording, passive constructions, and fragmented paragraphs.
- Reconstruct logic chains: replaces surface connectives such as “not only / but also” or “first / second / finally” with more stable academic relations such as causality, elaboration, and contrast.
- Unify terms and subjects: keeps core concepts consistent and fixes subject-deficient sentences so claims read as author-owned judgments.
- Clean formatting: merges very short paragraphs, and removes mechanical bolding, colon-led lists, metaphorical quotation marks, and emojis.
- Run dual audits: produces a Draft rewrite, performs two self-audit passes, and returns a Final rewrite plus Summary of changes.

Boundaries

It is best used for literature reviews, introductions, mechanism analyses, and discussion sections where the underlying argument is already complete. The goal is to reduce AI-sounding phrasing and assistant-like expressions, not to add experimental evidence, verify data, or invent citations. If the source contains incorrect references, missing context, or weak evidence, the rewrite may preserve those scholarly risks; manual verification of terms, data, and citations is still required before formal submission.