Applying Machine Learning and Natural Language Processing to Detect Phishing Email
Metadane
- Autorzy: A. Alhogail, A. Alsabih
- Rok: 2021
- Źródło: Computers & Security, Volume 110, 102414
- DOI/Link: 10.1016/j.cose.2021.102414
- Status: to-read
- Pochodzenie: Wyekstrahowane z al-subaiey-web-ai-phishing-2024 ([21] - GCN+NLP 98.2% accuracy, Table 3 comparison)
- Tagi: to-read reference graph-convolutional-network nlp phishing-email clair-dataset
Notatki
Publikacja dodana automatycznie z bibliografii.
Kontekst cytowania w Al-Subaiey 2024:
- Referenced jako [21] w paper (Table 3 comparison)
- Method: Graph Convolutional Network (GCN) + NLP techniques (tokenization, stop word removal)
- Dataset: CLAIR collection of fraud emails [34] - 8,579 emails (3,685 spam, 4,894 ham)
- Results: Accuracy 98.2%, False Positive Rate 0.015
- Comparison: Al-Subaiey 99.1% accuracy outperforms this baseline
Technical Details:
- GCN converts document classification → node classification problem
- Represents relations between entities as graph
- NLP preprocessing: tokenization + stop word removal
Significance:
- Novel application of GCN dla email classification
- Low false positive rate (0.015) - important for production
- CLAIR dataset became part of Al-Subaiey’s literature review context
Dodaj PDF aby wygenerować pełne podsumowanie używając /summarize-paper alhogail-gcn-phishing-email-2021