Evaluation on Malicious URL Detection with Different Features Based on Various Machine Learning Algorithms

Xiang Guo

2023

Abstract

With the escalating demand for cybersecurity, the identification of malevolent Uniform Resource Locators (URLs) has assumed paramount significance in defending against cyber threats. Various techniques, ranging from blacklists and heuristics to machine learning methods, have been employed for the purpose of detecting malicious URLs. Among these methodologies, machine learning stands out prominently due to its scalability, adaptability to emerging threats, and capacity to uncover threats that were hitherto unknown. This paper focuses on analyzing deep learning learning methods to detect malicious URLs compared with two traditional machine learning algorithm: Logistic Regression (LR) and K Nearest Neighbour (KNN). Scratching and collecting over 200,000 data to train the model and make prediction and evaluation. The result shows that the deep learning algorithm could achieve much higher scores than the other two machine learning models, but has much lower efficiency. The KNN model has better performance on selected feature group than hybrid feature group. The LR model could achieve higher performance on huge dataset and extremely complex feature group.

Paper Citation

in Harvard Style

Guo X. (2023). Evaluation on Malicious URL Detection with Different Features Based on Various Machine Learning Algorithms. In Proceedings of the 1st International Conference on Data Analysis and Machine Learning - Volume 1: DAML; ISBN 978-989-758-705-4, SciTePress, pages 338-342. DOI: 10.5220/0012808400003885

in Bibtex Style

@conference{daml23,
author={Xiang Guo},
title={Evaluation on Malicious URL Detection with Different Features Based on Various Machine Learning Algorithms},
booktitle={Proceedings of the 1st International Conference on Data Analysis and Machine Learning - Volume 1: DAML},
year={2023},
pages={338-342},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0012808400003885},
isbn={978-989-758-705-4},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 1st International Conference on Data Analysis and Machine Learning - Volume 1: DAML
TI - Evaluation on Malicious URL Detection with Different Features Based on Various Machine Learning Algorithms
SN - 978-989-758-705-4
AU - Guo X.
PY - 2023
SP - 338
EP - 342
DO - 10.5220/0012808400003885
PB - SciTePress