Change search
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
En fallstudie om upptäckt av nätfiske-e-post med stora språkmodeller
Jönköping University.
Jönköping University.
2025 (Swedish)Independent thesis Advanced level (degree of Master (One Year)), 10 credits / 15 HE creditsStudent thesisAlternative title
A Case Study on Phishing Email Detection with Large Language Models (English)
Abstract [en]

Phishing remains a critical cybersecurity threat, increasingly leveraging psychological manipulation and obfuscation to evade traditional detection. This study investigates whether DeepSeek-R1, the advanced large language model (LLM) currently open source, can be fine-tuned to be an effective solution for phishing email detection. Using public datasets from Hugging Face, the model was fine-tuned via the Unsloth framework and tested on both standard and obfuscated phishing samples. The model achieved an accuracy of 83.3%, a precision of 85%, and a recall of 80%, indicating good overall performance. It detected 89% of text-obfuscated and 96% of URL-obfuscated phishing emails, suggesting robust semantic understanding beyond keyword matching. However, challenges remain in minimizing false negatives and adapting to complex spear-phishing scenarios. This case study demonstrates the potential of fine-tuned LLMs for text-based phishing detection and highlights directions for future research in adversarial robustness and multimodal detection integration.

Place, publisher, year, edition, pages
2025. , p. 20
Keywords [en]
Cybersecurity, Phishing Detection, Large Language Models, Fine-Tuning, DeepSeek-R1, Obfuscation Techniques
National Category
Computer Sciences
Identifiers
URN: urn:nbn:se:hj:diva-69662OAI: oai:DiVA.org:hj-69662DiVA, id: diva2:1994429
Supervisors
Examiners
Available from: 2025-09-04 Created: 2025-09-02 Last updated: 2025-10-13Bibliographically approved

Open Access in DiVA

fulltext(780 kB)367 downloads
File information
File name FULLTEXT01.pdfFile size 780 kBChecksum SHA-512
da930c54de080c7c99418d73288382d436dc803b808ac64b44ae8944c31d59cbf181427d82d6e5542460690453f5dc9ac16adad209ad0e2ebb3e1b408b5130ea
Type fulltextMimetype application/pdf

By organisation
Jönköping University
Computer Sciences

Search outside of DiVA

GoogleGoogle Scholar
Total: 368 downloads
The number of downloads is the sum of all downloads of full texts. It may include eg previous versions that are now no longer available

urn-nbn

Altmetric score

urn-nbn
Total: 898 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf