Model of the Internet Traffic Filtering System to Ensure Safe Web Surfing

Abstract

The paper proposes a generalized model of network traffic filtering system that includes three modules: network operations’ module (low-level drivers needed to capture and modify network traffic); initialization and control module (initiates classification operations, preparing the received content for transfer to the classification module; manages the network operations’ module; keeps statistics on the detection of unwanted resources); web resource classification module (determines whether information provided in HTML format refers to one of three thematic categories that are banned for children). The system provides safe Internet surfing of the child and excludes access to prohibited categories with a probability of 99.53%. Classification is subject to HTTP request and HTTP response. Studies have been performed on the work of SVM classifiers and Naive Bayes classifier depending on the response threshold. The effect of stop words removal on classification accuracy of input data is also analyzed; it is able to increase the classification accuracy up to 6%.