Abstract
Phishing attacks are gaining sophistication for stealing sensitive information from a victim. The main idea behind this sophistication is the use of various psychological methods to influence a victim. This study focuses on the impact of linguistic cues of subjectivity and vagueness for detecting phishing emails. Vagueness is a useful differentiator because phishers highly rely on ambiguous wording than a legitimate email. Subjectivity is a strong indicator because phishing messages usually contain opinions, emotions, or exaggerated claims to manipulate users. Using a dataset, we identified the presence of these cues and incorporated them into machine learning models for phishing email classification. Experimental results show that the models leveraging linguistic cues of vagueness and subjectivity consistently outperform baseline classifiers in terms of F-score, demonstrating the effectiveness of such linguistic features in distinguishing phishing from legitimate emails. These findings highlight the potential of linguistically aware approaches for strengthening anti-phishing technologies.
Recommended Citation
Shovo, Md. Shafiq Newaj and Valecha, Rohit, "Phishing Email Detection Using Linguistic Cues" (2025). NEAIS 2025 Proceedings. 6.
https://aisel.aisnet.org/neais2025/6
Abstract Only