Despite the advancements made by researchers, spam emails remain one of the biggest challenges in the field of cybersecurity. Spam emails can serve as phishing emails or carry viruses that compromise the security of an organization's system. Current detection techniques depend on supervised learning or rely on cloud-based services, which can compromise user data privacy and affect implementation flexibility. This paper evaluates the capability of five large language models (LLMs) in zero-shot spam email classification. The models used in this study include llama3.1:8b, deepseek-r1:8b, gemma3:4b, falcon3:7b, and mistral:7b. In addition to predicting whether the email is spam or not, the LLM was also asked to generate an explanation of its prediction in natural language form. The experiments were conducted on two benchmark datasets: the Ling and TREC2007 datasets. In terms of performance, llama3.1:8b outperformed other LLMs when evaluated on the TREC2007 dataset (98.78% accuracy) and deepseek-r1:8b had the best performance on the Ling dataset (98.79%). The results show that open-weight LLMs can achieve competitive spam detection performance in a local, privacy-preserving environment without any fine-tuning.
Vusal Shahbazov· 2026 7th International Confe...· 0 citations
The growth of social media platforms results in billions of user-generated messages daily, making automated text analysis critical. The global impact of disinformation, the evolution of cyber threats toward psychological tactics, and the growing political and commercial value of public opinion have made sentiment analysis an important and relevant area of research. Although existing reviews have examined sentiment analysis approaches in terms of methods and application areas, few have evaluated these approaches in the context of cyber threat detection. This paper examines sentiment analysis methods applied to social media text data, covering lexicon-based, machine
learning, and deep learning approaches, including transformerbased architectures, as well as widely used datasets. The paper also discusses how sentiment analysis can be applied to the detection of
threats that exploit human emotions, including phishing, disinformation, and social engineering. To support this, an empirical analysis of large language model performance is conducted, measuring their ability to detect emotionally manipulative content. The purpose of this paper is to provide
readers with an objective understanding of sentiment analysis and its role as a defense against socially engineered cyber threats.
Vusal Shahbazov· “Kibertəhlükəsizlik və rəqəm...· 0 citations