In progress
Metadata analyzer
Python tool, with a console mode and a graphical interface, to analyze, extract, wipe and edit metadata in PDF, Word (.docx) and image files (JPG, PNG, HEIC and other Pillow formats). It includes an anomaly-detection model to flag files with suspicious patterns. It started as my Bachelor’s thesis.
What I did and how it works
- Metadata analysis and a check that the file extension matches its real (MIME) type, suggesting the right one if not.
- Metadata wiping for privacy, a default template and custom key=value fields. It always writes a new file and keeps the original.
- Anomaly detection with an Isolation Forest, to correlate suspicious patterns with potential threats.
Results
- My own estimate, not objectively measured: ~70 % less manual review time.
Tech stack
Usage details
- Still in development: next up are bulk processing and support for more formats.