Context-sensitive keyword density based supervised learning techniques for detection of malicious web pages

---

Tezin Türü: Yüksek Lisans

Tezin Yürütüldüğü Kurum: Orta Doğu Teknik Üniversitesi, Fen Bilimleri Enstitüsü, Fen Bilimleri Enstitüsü, Türkiye

Tezin Onay Tarihi: 2016

Öğrenci: BETÜL ALTAY

Danışman: AHMET COŞAR

Özet:

Conventional methods use a black list in order to decide whether a web page is malicious or not. These black lists are generally produced by technicians or operators and used for the security purposes of the organizations, protection of software from web based virus attacks, web browsers, etc. However, the blacklist approach is not a scalable solution for the frequently changing and rapidly growing number of web pages on the internet and their dynamic contents. In this thesis, we propose and analyze a method for the classi cation of the web pages by using Support Vector Machine, Maximum Entropy, and Extreme Learning Machine techniques. The performance of the proposed machine learning models are evaluated with 100K web pages. Features of web pages are generated by processing HTML contents and information is obtained using conventional feature extraction methodologies, such as existence of words, keyword frequencies, and a novel method based on keyword densities. The performances of machine learning methods employing various extracted features are analyzed and experimental results show that the proposed method can identify malicious web pages with a very high accuracy of up to 98.24% while also achieving practical web page processing times.