Adaptive Online Retail Web Site Based on Hidden Markov Model

Abstract

There are two problems in online retail: 1) The different interest of all customers in the different commodities conflicts with the commodity classification structure of the web site; 2) Customers will simultaneously buy some commodities that are classified in different classes and levels in the web site. The two problems will make the majority of customers access overabundant web pages. To solve these problems, we mine the web data to build a hidden markov model, use association rule discovery to get the large item sets, use Viterbi algorithm to find some paths, mark the large item sets in the nodes of the paths. The large item sets will compete in the nodes for the limited space. Through this approach the web site will adjust itself to reduce the total access times of all users. This method also can be used in analyzing paths, advertisements, and reconstructing the web site.