In this work, we compare various text-based pornographic Web filtering techniques. The techniques include blacklist and keyword blocking. The technique called SV is modified to extract a representative feature vector. Each test Web pagepsilas feature is extracted and gathered as a vector. The vector is then summarized and compared with the global representative vector to justify whether the Web page is pornographic or not. The experiments tested on the traces of the log file show that the representative-based algorithm performed well in overall compared with keyword blocking algorithm.