Semester of Graduation

Fall 2024

Degree

Master of Science in Computer Science (MSCS)

Department

Computer Science and Engineering

Document Type

Thesis

Abstract

Large Language Models (LLMs) excel in diverse natural language tasks but often lack specialization in fields like digital forensics. Their reliance on cloud-based APIs or high-performance computers restricts their use in resource-limited environments, and response hallucinations could compromise their applicability in forensic contexts. We introduce ForensicLLM, a 4-bit quantized LLaMA-3.1-8B model fine-tuned using Retrieval Augmented Fine-tuning (RAFT) on 6,739 Q&A samples extracted from 1,082 digital forensic research articles and curated digital artifacts. Evaluation on 2,244 Q&A samples showed that ForensicLLM outperformed the base LLaMA-3.1-8B model by 4.06%, 5.43%, and 15.79% on BERTScore F1, BGE-M3 cosine similarity, and G-Eval, respectively. Compared to LLaMA-3.1-8B + RAG, it achieved improvements of 3.46%, 3.25%, and 4.61% on the same metrics. ForensicLLM accurately attributes sources 86.6% of the time, with 81.2% of the responses including both authors and titles. Additionally, a user survey conducted with digital forensics professionals confirmed significant improvements of ForensicLLM and the RAG model over the base model across multiple evaluation metrics. Participants engaged with hypothetical scenarios, with ForensicLLM showing strength in correctness and relevance metrics, while the RAG model was appreciated for providing more detailed responses. These advancements mark ForensicLLM as a transformative tool in digital forensics, enhancing model performance and source attribution in critical investigative contexts.

Date

11-17-2024

Committee Chair

Baggili, Ibrahim

Available for download on Saturday, November 01, 2025

Share

COinS