Abstract:
"The growth of pornography on the internet had created a significant challenge for parents and others worried about exposing their children to graphic sexual content. Porn detection
algorithms were developed to identify and limit access to pornography websites and other
sexually explicit internet material. However, these technologies had significant limitations and frequently failed to block explicit content or non-pornographic websites. Furthermore, porn detection software could be circumvented by accessing prohibited content using proxy servers, VPNs, or other means. As a result, a more effective and comprehensive strategy for identifying and blocking explicit information on the internet was required.
The goal of this research project was to create, construct, and test a hybrid identification model capable of identifying and blocking explicit material on the client side, with a primary focus on pornographic content as explicit content. By combining several elements of websites to identify and filter explicit material, the developed model addressed the limitations of existing detection systems. The model utilized a combination of image analysis and text analysis to accurately detect and block explicit content on the web, employing a CNN for image classification with an accuracy of 89.49% and an NLP LSTM model for text classification with an accuracy of 97.06%.
This research effort resulted in a complete hybrid identification model capable of properly
detecting and blocking explicit content on the internet. This strategy assisted parents, schools, and internet service providers in safeguarding their children from inappropriate information. The findings of this study contributed to creating more effective measures for promoting internet safety and protecting individuals, particularly children, from the detrimental impacts of explicit information on the internet.
"
