WebGuard: A Web Filtering Engine Combining Textual, Structural, and Visual ContentBased Analysis

1
WebGuard A Web Filtering Engine Combining
Textual, Structural, and Visual Content-Based
Analysis

• Mohamed Hammami, Youssef Chahir, and Liming Chen

Presented by Xin Hu
2
INTRODUCTION

• Filtering Undesirable Web
• WebGuard An Automatic Machine Learning-Based
  Pornographic Website Classification and Filtering
  System
• Combination of Textual, Structural Content-Based
  Analysis and Skin Color Related Visual
  Content-Based Analysis

3
Related Research

• Platform for Internet Content Selection (PICS)
• URL Blocking
• Keyword Filtering
• Intelligent Content-Based Analysis

4
Commercial Filtering Products

• Microsoft Internet Explorer (RSACi)
• Cybersitter
• Netnanny
• Norton Internet Security
• Puresight Home
• Cyber Patrol

5
Classification Accuracy
6
Principle and Architecture

• Textual content-based features
• Structure content-based features
• Visual content-based analysis skin color-like
  pixel classification

7
Skin Color Modeling

• Supervised Learning Process
• Region Growing-Based Image Segmentation

8
Image Processing Results
9
Experimental Results
Classification Accuracy Rate WebGuard and
Commercial Products
10
Conclusion

• Integration of textual, structural content-based
  analysis and skin color related visual
  content-based analysis
• Achieved 97.4 classification accuracy, compared
  with less than 90 for commercial products

11
Questions and Future Work

• Application of this technique on video?
• The speed of navigation delayed significantly by
  the filtering engine?
• Integration of other image processing techniques,
  like pattern recognition?
• More