Research on Application of Clustering Algorithm Based on PSO for the Web Usage Pattern

A new clustering method, RVPSO-K was proposed in this paper based on the random research combining K-means and PSO according to the character of the clustering of Web usage pattern and the disadvantages of particle swarm optimization. The searching ability of the individual particle was enhanced by change the flying trajectory of the particle, hence, the optimum would be found in a relatively short amount of time. The experiments were implemented on Web log data after data preprocessing and showing that the RVPSO-K is better than the K-means based on PSO clustering algorithm in the aspects of the stability, the precision and convergence speed. In addition, a different threshold was proposed; therefore the agility of the recommended Web was improved to satisfy distinct requirements.