Title: A pornographic web page detecting method based on SVM model using text and image features

Authors: Rung-Ching Chen, Chun-Te Ho

Addresses: Department of Information Management, Chaoyang University of Technology, 168 Gifeng E.Rd., Wufeng, Taichung Country, Taiwan, ROC. ' Department of Information Management, Chaoyang University of Technology, 168 Gifeng E.Rd., Wufeng, Taichung Country, Taiwan, ROC

Abstract: In this paper, a new web page filtering method directed at pornographic material and based on image and text content analysis is proposed. First, the features of image and text are extracted from the web page content. An analysis is performed and the results are merged. The page is then rated pornographic or non-pornographic. To facilitate real-time analysis, we also propose an acceleration method. In experimental results, text classification accuracy is 95.8% and image classification accuracy is 84%. In addition, the accuracy of web page classification after merging analysis of text and image features is 91.8%.

Keywords: support vector machines; image classification; text classification; pornography; web pages; filters.

DOI: 10.1504/IJIPT.2006.010561

International Journal of Internet Protocol Technology, 2006 Vol.1 No.4, pp.264 - 270

Published online: 01 Aug 2006 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article