The performance of Support Vector Machine is higher than traditional algorithms. The training process of SVM is sensitive to the outliers in the training set. Here in this Paper, a new approach called, Web Pages Categorization based on Classification and Outlier Analysis (WPC-COA), is proposed that uses a polynomial Kernel function to map web page tuples to high dimensional feature space.
This study uses a new methodology which helps in mapping web page tuples with various attributes such as frequency, time spent on each page, in-degree, out-degree and level of a web page to high dimensional feature space. The paper’s primary contribution is to categorize web pages based on classification and outlier analysis using Polynomial Kernel function.
J. C. Platt, Sequential minimal optimization-A fast algorithm for training support vector machines, in advances in kernel methods-support vector learning. Cambridge, MA: MIT Press, 1998.
D. Gomez, J. Montero, and G. Biging, "Improvements to remote sensing using fuzzy classification, graphs and accuracy statistics," Pure Appl. Geophys., vol. 165, pp. 1555-1575, 2008.
E. O. Edgar, F. Robert, and G. Federico, Support vector machines: Training and applications: A.I. Memo No. 1602, C.B.C.L Paper No.144, March 1997, 2004.
P. Scott and H. Lutz, "Comparing the results of support vector machines with traditional data mining algorithms, supported by Amica Life Insurance Company."
W. Xiao-Hong, "Coll. of electr and inf eng, Jiangsu Univ, Zhenjiang. A possibilistic C- means clustering algorithm based on kernel methods," presented at the Computational Intelligence for Modelling, Control and Automation, 2005 and International Conference on Intelligent Agents, Web Technologies and Internet Commerce, Nov 2005.