ANALISIS MALICIOUS URL PADA FILE MENGGUNAKAN METODE K-MEANS CLUSTERING BERBASIS HOST-BASED FEATURE EXTRACTION
Total View This Week0
Total View This Week0
Institusion
Universitas Sriwijaya
Author
RAFI, MUHAMMAD IMAM
Heryanto, Ahmad
Heryanto, Ahmad
Subject
T1-995 Technology (General)
Datestamp
2023-05-12 05:06:25
Abstract :
The attacks or threats faced by internet users today have various types of attacks. These attacks are attacks in the form of phishing, malware, spyware, and ransomware. One of the most effective means of cyber attack carried out by attackers is by using URLs. URL (Uniform Resource Locator) is an address used to find the location of a file on the Internet. This makes URLs used as a method for carrying out cyber attacks referred to as Malicious URLs. Malicious URLs or dangerous sites on the internet contain a lot of content in the form of spam, phishing, which is used to initiate attacks. In this study, generate a URL dataset with URL features in the form of DNS records from URLs that will be used as data in clustering with K - Means. And produce a visualization of data clustering results with K - Means using a value of k = 2, namely in the form of benign clusters and malicious URLs. And analyze the visualization results of clustering with K - Means using the clustering validation test using the Silhouette Score with a result of 72.62% for k=2. In this study, generate model validation by training the URL dataset on machine learning and applying Hypeparameter tuning so that the performance results for each cluster are benign (0) 85% precision, 97% Recall, 91% F1-Score, and malicious (1) precision clusters. 96%, Recall 83%, F1-score 89%, and the accuracy of the model used is 89.94%.
The attacks or threats faced by internet users today have various types of attacks. These attacks are attacks in the form of phishing, malware, spyware, and ransomware. One of the most effective means of cyber attack carried out by attackers is by using URLs. URL (Uniform Resource Locator) is an address used to find the location of a file on the Internet. This makes URLs used as a method for carrying out cyber attacks referred to as Malicious URLs. Malicious URLs or dangerous sites on the internet contain a lot of content in the form of spam, phishing, which is used to initiate attacks. In this study, generate a URL dataset with URL features in the form of DNS records from URLs that will be used as data in clustering with K - Means. And produce a visualization of data clustering results with K - Means using a value of k = 2, namely in the form of benign clusters and malicious URLs. And analyze the visualization results of clustering with K - Means using the clustering validation test using the Silhouette Score with a result of 72.62% for k=2. In this study, generate model validation by training the URL dataset on machine learning and applying Hypeparameter tuning so that the performance results for each cluster are benign (0) 85% precision, 97% Recall, 91% F1-Score, and malicious (1) precision clusters. 96%, Recall 83%, F1-score 89%, and the accuracy of the model used is 89.94%.
File :
RAMA_56201_09011281823065.pdf
RAMA_56201_09011281823065_TURNITIN.pdf
RAMA_56201_09011281823065_0022018703_01_front_ref.pdf
RAMA_56201_09011281823065_0022018703_02.pdf
RAMA_56201_09011281823065_0022018703_03.pdf
RAMA_56201_09011281823065_0022018703_04.pdf
RAMA_56201_09011281823065_0022018703_05.pdf
RAMA_56201_09011281823065_0022018703_06_ref.pdf
RAMA_56201_09011281823065_0022018703_07_lamp.pdf
Download
book
BibTex
Latex, Jabref
cloud_download
Original Resource
url resource Institution
