Multi level filtering to classify and block undesirable explicit material in website

Abstract

The growth of Internet opens wide opportunities for content production in various types of internet sites that provide information of various categories. For students or kids, ease of access to the internet site have a positive impact to provide days connectivity of user to the content that they needed for their study, but the other side, it also cause a negative impact when they accessing sites that provide undesirable explicit material such as pornographic content. During this time, pornographic content can be easy inserted through the text or images in page of internet sites. Therefore, we need a system that can filter information related to pornography and can prevent users who are not old enough to access pornographic content. To overcome this issue, we develop an online client-side filtering system that allows the user to perform a domain filtering, URL filtering, keyword filtering, and skin detection filtering. Also we proposed a hybrid skin color detection technique to overcome the failure of detecting skin when the images in close-up or scene mode. This system consist of classification using n-gram tokenizer for text and haar algorithm and skin chromaticity at HSV (Hue Saturation Value) and Normalized RGB color spaces for images. The applications created using software development tools such as flowchart and Unified Modelling Language (UML), while for the programming language using PHP, CSS, Javascript and using MySQL database.