Publications

Abstract : Be it formal or casual, email is undoubtedly the most popular means of communication in modern times. Their popularity owes to the fact that they are reliable, fast and more over free to use. One issue that plagues this otherwise solid technology is phishing emails received by users. Phishing emails have always bothered users as it's a huge waste of storage, time, money and resource to any user. Many previous attempts to eradicate or at least block phishing emails have been deemed futile. This work uses word embedding as text representation for supervised classification approach to identify phishing emails. Ruled based and machine learning models with feature engineering were attempted but failed due to the ever increasing ways of threats and lack of scalability of the model. Deep learning based models have shown to surpass the older techniques in spam email detection. This work aims at attempting the same using a CNN/RNN/MLP network with Word2vec embeddings on phishing email corpus, where Word2vec helps to capture the synaptic and semantic similarity of phishing and legitimate emails in an email corpus. This work aims to show the abilities of word embedding have to solve issues related to cybersecurity use cases. Copyright © by the paper's authors.