Web traffic and e-commerce activities are increasing rapidly; hence, understanding the behavior of users based on their interactions with websites is becoming more and more important. To do so, web usage mining is needed. This research analyzes web clickstream data to extract usage patterns. There are two major challenges involved in Web usage mining. The first one is preprocessing the raw data to provide an accurate picture of how a website is used. The second one is to present the rules and patterns that are potentially interesting to the users by filtering the results. This forms the basis for this thesis, where a novel real-time system is discussed. This system builds personalized browsing assistance based on website user request(s) submitted to the web server and past user(s) behavior. Our proposed system is of crucial importance to users browsing the internet. Providing accurate link suggestions is one of the advantages of the system. This has been further developed to a live screenshot of the suggested web page. This enables the user to preview the content before making visiting the web page. Besides this, the proposed system can provide suggestions based on the user’s browser and operating system. This means that every browser and operating system has a unique suggestion model customized to its user. To evaluate the system, we provide a user study, case studies and conduct experiments on five datasets to verify the effectiveness of our proposed system.
Keywords
Real-Time AnalyticsConsumer BehaviorCatBoostDiscrete Time Markov ChainPRISM
Full Study
Institute(s)
University of Alberta
Year
2022
Abstract
Author(s)
Syed Tauhid Zuhori