GOOGLE PAGERANK MODEL EXPLORATION (CASE STUDY: BANDUNG INSTITUTE OF TECHNOLOGY WEBPAGE)

• Main Author: Rizki Fadillah, Muhammad
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/41326
• Institution: Institut Teknologi Bandung

In this thesis, the author studied about a method to rank webpages with a model called PageRank. The model is proposed by the founders of Google search engine, Larry Page and Sergey Brin. This model ranks webpages according to the link structure within. By using the model, we would know which webpage will be most visited if we surf on the webpages by the links continuously. On its development, a teleportation vector which enabling personalization of the page rankings according to user’s preferences was introduced. This model was developed into a model with time-dependent (dynamic) teleportation vector. Dynamic PageRank model is a differential equation that involving time variable, teleportation vector and transition matrix. Output of this model is a vector that indicate tendencies of the webpages to be accessed which are time-dependent. In this thesis, both classic (without personalization) and dynamic PageRank model will be applied to Bandung Institute of Technology webpage. Classic PageRank model will be run with the power method and the PageRank value obtained will be used as an initial value for the dynamic PageRank model. Dynamic PageRank model will be evaluated using Euler method (to obtain the numerical solution of the differential equation) enhanced by Bandung Institute of Technology webpages visit data as an estimator for teleportation vector henceforth linkages between the estimator and model results will be identified. The results of this research are the homepage (https://www.itb.ac.id/ ) is the number third of most significant pages list according to classic PageRank model, and the estimator data affect the dynamic model directly proportional but with a little time lag.