B1
Detail B1 · Code similarity through structural explorationCompleted

Finding the fingerprint in a program

Two programs can look completely different and still share the same structure. Renaming variables or reformatting code fools text-based comparison, but not structural analysis.

Date
Sep 2025 – May 2026
Status
Completed
Team
6-person team
Institution
Arts et Métiers
Location
Metz, France
01

Context

Applications range from detecting plagiarism in programming assignments to recognizing the patterns and watermarks that generative AI leaves in code.

02

Objectives & constraints

Build a program that measures the similarity between Python programs from their structure rather than their text.

03

Team & role

Alain ÉtienneSupervisor · Arts et Métiers, Metz
Rafael BratkowskiTeammate · Arts et Métiers, Metz
Gabriel Carvalho MendesTeammate · Arts et Métiers, Metz
Ewenn FillonneauTeammate · Arts et Métiers, Metz
Evan JouaultTeammate · Arts et Métiers, Metz
Mathias LacpatiaTeammate · Arts et Métiers, Metz
BOM

Tools & technologies

Python
Program analysisAI detection
DWG NO. WH-2026-001 · William Hochart · Arts et Métiers Institute of TechnologyAll projects · CV · © 2026