| Abstract [eng] |
Although the internet is considered as one of the most significant technological advancements, attention is being drawn to its negative aspects, such as access to harmful or unwanted content by minors and other users. During this study, a machine learning-based website filtering system for routers was developed. The process of categorizing websites is performed on a central server, while filtering is done on the router using DNS blocking. To expedite query processing on the router and reduce resource usage, caching was employed. Additionally, firewall configuration rules were created to prevent circumvention of the website filtering system's content blocking. Research was conducted to assess the impact of the quality of the training data set on the system's performance, evaluate the feasibility of deploying the developed system or its components in an embedded environment, examine the effectiveness of machine learning algorithms used for text processing in categorizing websites, and assess the accuracy and performance of the machine learning algorithms and their combinations used in the website categorization system. |
