A METHOD OF PORNOGRAPHIC WEBSITE DETECTION BASED ON META TAGS AND DEEP LEARNING

In order to solve the problem that pornographic websites are increasingly disguised and hidden, the text content of pornographic websites is analyzed. The current research focuses more on images and ignores the fine granularity analysis of text, and does not effectively use text information to detect pornographic websites. In order to solve the above problems, the HTML source code was drilled down, a method of pornographic website detection based on meta tags was proposed, and a mass pornographic website detection model (MPWM) was constructed. Compared with the benchmark model, MPWM model achieved the best detection effect in accuracy, recall and F1 value, and F1 value has reached 97.62%.