A comparison of Bayesian networks learning algorithms: A case study of webpage layout design

Due to the Internet is become prevalent, webpage is a vital medium, which does duty as a channel in communication process between a website-designer and users. Webpage design consists of many elements such as color, image, layout etc., which may contribute to users perception. The authors have created a knowledge model of webpage layout design by using Bayesian network technique. In order to build appropriate models, several algorithms for learning Bayesian networks have been developed. In this study, four knowledge models of webpage which created based on different structure learning algorithm of Bayesian network were compared for finding the most appropriate algorithm. As a result, it becomes clear that the model created with Tabu search and the K2 algorithm is the most appropriate.