Evaluasi Efektivitas Penggunaan FastText Embedding dan LSTM Networks dalam Deteksi Phishing Email

Rukiman, Sheptianna Healtha (Informatics Department, Faculty of Engineering, Siliwangi University)
Rahmatulloh, Alam (Program Studi Informatika, Fakultas Teknik, Universitas Siliwangi)

Faktor Exacta
Vol 18, No 2 (2025)

Phishing emails represent a significant cyber threat, necessitating advanced detection methods. This study evaluates a model combining FastText word embedding and a Long Short-Term Memory (LSTM) neural network to identify these threats. Using a public dataset from Kaggle, the model was trained on 80% of the data and tested on the remaining 20%. The methodology included data preprocessing, vectorization with FastText to capture sub-word information, and sequential pattern recognition using the LSTM architecture. Performance was evaluated using accuracy, precision, recall, and F1-Score, with the model achieving a 92% detection accuracy. Key challenges identified include class imbalance and high computational requirements. Future research could focus on model optimization and data augmentation techniques to further enhance detection performance and address these limitations.