A Malicious File Checker for FKI Report Submission using Supervised Learning

Abstract

This paper presents the idea of applying a machine learning algorithm as a classifier for a malicious PDF file checker. PDF files might seem harmless but the truth is these files may contain objects such as JavaScript code or binary code that may be malicious. Therefore, an academic institute should have a malicious file checker in making sure that the file submitted to the institute’s website does not contain any malware that would corrupt the system. Hence, in this study, the aim is to investigate machine learning algorithms that could be applied for malicious file checkers and to design and evaluate the malicious file checker in terms of its accuracy in finding malware inside the PDF files. Since the malicious file checker is machine learning-based, some machine learning algorithms first must be put into an evaluation to conclude which algorithm has a better score in making predictions to be picked as the classifier. Next, the machine learning algorithm will then be integrated into the malicious file checker system and some data set will be tested to get the perfect score in predicting whether the file uploaded is malicious or not. A higher score is to be estimated for the built of a malicious file checker.