AI-Enhanced Forensic Tool for macOS: Semantic Vector Search and Regex-Based Risk Detection

This repository provides the source code for an academic forensic tool developed as part of a Master’s thesis in Informatics (Privacy, Information, and Cybersecurity) at the University of Skövde. The tool is designed to enhance digital forensic investigations on macOS systems by leveraging semantic vector search and deterministic regex-based pattern detection. It integrates: SentenceTransformer embeddings (all-MiniLM-L6-v2) to convert file content into high-dimensional semantic vectors. Oracle 23ai for Approximate Nearest Neighbor (ANN) search using cosine similarity. Regex scanning for sensitive entity detection (e.g., passwords, emails, IPs, credit card numbers). Visual risk classification and reporting via Python. The system detects renamed, obfuscated, and semantically hidden sensitive files across .txt, .pdf, .csv, and .json formats. While Large Language Models (LLMs) are discussed, the implementation uses no active LLM inference, ensuring forensic reproducibility and transparency.