Year: 2025
Type: conference-paper
Abstract: Digital investigators face challenges searching through massive data when coded language evades keyword searches. Our research shows how vector embeddings enhance digital forensics by detecting implicit drug references in social media. Using LLM2Vec with instruction-based embeddings, we significantly outperformed traditional keyword searches when dealing with slang, enabling investigators to find content based on meaning rather than exact word matching.
Authors: Adil Koeken, Samay Nandwana, Avantika Shah, Vinicius Lima, Umit Karabiyik
Institution: Purdue University West Lafayette
Language: English
Topic: Digital and Cyber Forensics
Subfield: Information Systems
Field: Computer Science
Domain: Physical Sciences
Open Access status: gold
OpenAlex ID: W4410220988