Style Scalpel Feature Graph Map

This repository hosts the interactive feature-profile graphs for the Style Scalpel stylometric detector project.

The graphs visualize normalized line profiles for 132 stylometric features, separated into AI-generated and human-written samples.

Open the graphs

Main page:

https://m0333ism-dev.github.io/Features-Graph-Map/

Direct graph pages:

• AI feature graph: https://m0333ism-dev.github.io/Features-Graph-Map/ai_features_lines.html
• Human feature graph: https://m0333ism-dev.github.io/Features-Graph-Map/human_features_lines.html

Repository contents

• index.html — landing page for opening the graph files
• ai_features_lines.html — interactive normalized feature-line graph for AI-generated samples
• human_features_lines.html — interactive normalized feature-line graph for human-written samples
• map_report.txt — short technical summary of the generated graphs
• .nojekyll — prevents GitHub Pages from processing the files through Jekyll

Data summary

The graphs are based on the PG 132-feature stylometric feature set.

Summary:

• Numeric features plotted: 132
• Normalization: min-max normalization to [0, 1]
• Human samples: 2,523
• AI samples: 2,523
• Constant/all-NaN feature handled: dep_cop_ratio was set to 0.0

The normalization was applied per feature across all samples, so the graph is meant for visual comparison of feature-profile shape, not for reading raw feature values.

Main Style Scalpel project

The main detector repository is available here:

https://github.com/m0333ism-dev/Style-Scalpel

Purpose

This repository is for research visualization. It shows how the AI and human sample groups behave across the 132-feature stylometric space.

It is not a standalone detector and should not be used by itself to judge whether a specific text is AI-generated or human-written.

Note

These graphs are exploratory visual materials connected to the Style Scalpel project. They are intended to support interpretation, comparison, and research presentation.

License

The visual graph materials in this repository are licensed under the Creative Commons Attribution 4.0 International License (CC BY 4.0).

You may share and adapt the graphs, provided appropriate credit is given to the Style Scalpel project.

Suggested attribution:

Muhammad Ismail. Style Scalpel Feature Graph Map. GitHub, 2026. Licensed under CC BY 4.0.