Style Scalpel research guide
How Style Scalpel Analyzes AI Authorship
See how Style Scalpel processes academic text, evaluates stylometric evidence, and presents paragraph-level and document-level AI authorship signals.
1. Supply text or a document
Paste formal prose or upload a supported TXT, DOCX, or PDF file. The web detector accepts files up to 8 MB and extracts the text required for analysis.
2. Process the writing
The application separates usable text into passages and prepares it for consistent analysis. PDF results can be affected by extraction quality, unusual layouts, scanned pages, or encoding problems.
3. Extract stylometric evidence
The detector measures a collection of linguistic and writing-style features. These provide evidence across lexical, syntactic, structural, punctuation, and related patterns rather than relying on one isolated clue.
4. Evaluate authorship patterns
The frozen research model evaluates the combined feature pattern. This produces model outputs that indicate how closely the analyzed writing resembles the conditions learned during training.
5. Review document and paragraph evidence
Style Scalpel presents an overall assessment alongside paragraph-level probabilities and influential style features. The detailed view supports human interpretation and can expose variation that a single score would conceal.
6. Generate a forensic-style report
An HTML evidence report can be opened or downloaded for inspection. Reports are temporary and are intended to support review, not to certify or prove authorship.