Comparison of PageRank Algorithm Implementations on a Single Computer

Abstract

Pagerank Algorithm is an algorithm used for calculating web page ranking in Google search engine. Problem arises for Pagerank Algorithm due to big main memory usage, thus make it impossible to run in single machine computer with limited main memory. Alternative algorithms will be proposed by comparing the alternative algorithms from other studies with the Original Google Pagerank in terms of speed, main memory usage, and their result similarity. In this study, the Orignal Pagerank, Distributed Pagerank Computation (DPC), Modified DPC, and Random Walker algorithms will be implemented. The implemented algorithms will be run with datasets, and their speed, main memory usage, and result similarity will be compared. For result similarity, Random Walker’s result will be used as a benchmark, since it has been used as base concept of Pagerank. It is concluded that the Original Pagerank is faster and has very similar result with Random Walker, while DPC and MDPC have significantly smaller main memory usage, thus very suitable for single machine computer with limited main memory, but run slower and sacrificing result similarity.