Clustering Websites Using a MapReduce Programming Model

In this paper, we describe an effective method of using Self-Organizing Map (SOM) to group websites so as to eliminate or at least ease up slow speed, one of the fundamental problems, by using a MapReduce programming model. The proposed MapReduce SOM algorithm has been successfully applied to cluB, which is a typical SOM tool. Performance evaluation shows the proposed SOM algorithm took less time to complete computational processing (i.e. distributed computing) on large data sets in comparison with conventional algorithms, and performance improved by up to 20 percent with increasing nodes (computers).