An intelligent categorization engine for bilingual web content filtering

• Published in: IEEE transactions on multimedia Vol. 7; no. 6; pp. 1183 - 1190
• Main Authors: Lee, P.Y., Hui, S.C., Fong, A.C.M.
• Format: Journal Article
• Language: English
• Published: New York, NY IEEE 01.12.2005; Institute of Electrical and Electronic Engineers; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Applied sciences; Artificial intelligence; Computer science; control theory; systems; Connectionism. Neural networks; Engines; Exact sciences and technology; Filtering; Filtration; Information filtering; Information filters; Intelligent networks; Internet; Messages; Multimedia; neural network applications; Neural networks; On-line systems; Performance analysis; Protection; Search engines; Speech and sound recognition and synthesis. Linguistics; Telecommunications; Telecommunications and information theory; Teleprocessing networks. Isdn; text processing; unsupervised learning; Web pages; Websites; World Wide Web; Word processing; English; text processing; Multilingualism; neural network applications; World wide web; Chinese; Internet; Neural network; Categorization; Unsupervised learning; Content analysis
• ISSN: 1520-9210; 1941-0077
• DOI: 10.1109/TMM.2005.858414

It is important to protect children and unsuspecting adults from the harmful effects of objectionable materials, such as pornography, violence, and hate messages, which are now prevalent on the World-Wide Web. This calls for effective tools for web content analysis and filtering of objectionable contents. Our study of existing web content filtering systems has identified a number of deficiencies in these systems. Using the analysis of pornographic web pages as a case study, we present an intelligent bilingual web page categorization engine that can determine if an English or Chinese language web page contains pornographic materials. We have implemented the categorization engine to perform offline web page analysis and near-instantaneous online filtering. Performance evaluation of our system has verified its effectiveness.