TOWARD BETTER WEBSITE USAGE: LEVERAGING DATA MINING TECHNIQUES AND ROUGH SET LEARNING TO CONSTRUCT BETTER-TO-USE WEBSITES

 
 
 
 

TOWARD BETTER WEBSITE USAGE: LEVERAGING DATA MINING TECHNIQUES AND ROUGH SET LEARNING TO CONSTRUCT BETTER-TO-USE WEBSITES

Show full item record


Title: TOWARD BETTER WEBSITE USAGE: LEVERAGING DATA MINING TECHNIQUES AND ROUGH SET LEARNING TO CONSTRUCT BETTER-TO-USE WEBSITES
Author: Khasawneh, Natheer Yousef
Description: When users browse a website, they usually try to accomplish a certain task, such as finding information, buying products, registering for classes, and attending classes on-line. The interaction between the users and the website can give the web engineers insight into the most common user tasks performed on the website. They can learn how most users navigate the website to finish their tasks and what changes can be made to the website structure in order to make the completion of the common tasks easier and faster. Most web servers provide web interaction logs to track the interaction between the users and the website. But such logs are usually designed for debugging purposes and not for the analysis of the website. So there is a need for a deeper conceptual method to analyze the interaction log to reveal information that can be used for enhancing the website structure. In this work, different data mining techniques, along with a rough set learning approach, are presented to enhance website usage. A new active-user-based user identification algorithm was applied to the interaction log to group together records that belong to the same user. The algorithm has a complexity running time of one order faster than other user identification algorithms. Sessions for identified users are found using an ontology-based session identification algorithm, which uses the website ontology in determining the sessions within website users. Different website sessions are then compared using a new Multidimensional Session Comparison Method (MSCM). MSCM takes into consideration other session dimensions, such as pages visited, time spent on the pages and the session length. MSCM compares sessions more precisely than other well known session comparison methods, such as the Sequence Alignment Method (SAM), Multidimensional Sequence Alignment Method (MDSAM), and Path Feature Space. Using the comparison results from the MSCM, sessions are clustered by hierarchal and equivalence classes clustering algorithms. The clustering results are used by the rough set learning method and the centroid method to generate rules that are used for both predicting and describing sessions’ clusters. Rules generated using a rough set learning approach predict and describe clusters better than rules generated using centroid method. Each session cluster is considered one task and the cluster centroid is the navigation path for completing the task. So common tasks along with their navigation path are evaluated, suggestions are then made for the website engineer to enhance the website structure to better serve website users. This work shows how data mining techniques along with rough set learning methods can be used to enhance the website structure for better-to-use websites.
Permanent Link: http://rave.ohiolink.edu/etdc/view?acc_num=akron1120534472
http://hdl.handle.net/2374.OX/3587
Date: 2005

Files in this item

Files Size Format View

There are no files associated with this item.

This item appears in the following Collection(s)

Show full item record