A Double Head Clustering Algorithm for Web Usage Mining Based on Radical Basis Function Neural Network

Abstract

Objectives: The main intention of this work is to propose dynamic recommended system which has a better understanding of navigation preferences of the online users and effectiveness of a website Methods: Our proposed method introduces an advance clustering architecture by introducing an improved cluster head selection mechanism with the effect of data elements similarity patterns. To distribute the similarity of data, two cluster head called primary and secondary cluster heads and both will be activated at the same time. The output patterns of each user profile are trained using Radical Basis Function (RBF) neural network Findings: The proposed method is compared with various traditional clustering approaches like an ant clustering, k-means clustering and spherical k-means clustering. Proposed system provides better quality when compared to the traditional clustering approaches. Improvements: The quality of the proposed system is evaluated in terms of precision, coverage and F1 measure. The proposed method is compared with various traditional clustering approaches like ant clustering, k-means clustering. The experiment results show that our proposed system provides better quality when compared to the traditional clustering approaches. Application: site improvement, site modification, business intelligence and usage characterization.