Usage
  • 7 views
  • 4 downloads

Online User Purchasing Behavior Modeling

  • Author / Creator
    Gong, Xiaohui
  • With the proliferation of e-commerce business, the study of online user purchasing behavior plays an important role in improving purchasing experiences of users as well as providing valuable intelligence to sellers. While most previous research efforts focused on explicit user behavior modeling, implicit user behavior modeling provides a greater amount of information and is more feasible and reliable for online purchasing scenario. In addition, the recommendation based on similarities in previous research results in biased, delayed or incorrect recommendation due to the absence of explicit multi-attribute modeling. Although some works have used multi-criteria decision making to solve this problem, the cardinal functions used have the attributes independence restriction that causes convex hull problem in online purchasing scenario. This thesis proposes a probabilistic multi-criteria item ranking framework that predicts the probability of an item being a user’s best choice and ranks items accordingly. It uses indifference curve in microeconomics to ordinarily model implicit user behavior by using users’ purchasing history directly. The newly designed ordinal model offers a flexible way to model any kind of user behavior with explicit multi-attribute modeling and without information bias/loss or convex hull problem. The model also considers inter-item competition globally. In addition, different from all prior works in which users are assumed to be able to compare all items simultaneously, the proposed prediction framework considers the fact that a user can only compare a few items at the same time, and models the user’s decision process as a two-step selection process, where the user first selects a few candidates, and then makes detailed comparison. Furthermore, according to the comprehensive simulation and real user test results, the proposed algorithm significantly outperforms existing multi-criteria ranking algorithms by achieving higher ranking accuracy with short learning curve. Besides making recommendations, the proposed framework in this thesis can further benefit online sellers to improve their marketing strategies.

  • Subjects / Keywords
  • Graduation date
    2017-11:Fall 2017
  • Type of Item
    Thesis
  • Degree
    Master of Science
  • DOI
    https://doi.org/10.7939/R3ZP3WD8K
  • License
    This thesis is made available by the University of Alberta Libraries with permission of the copyright owner solely for non-commercial purposes. This thesis, or any portion thereof, may not otherwise be copied or reproduced without the written consent of the copyright owner, except to the extent permitted by Canadian copyright law.
  • Language
    English
  • Institution
    University of Alberta
  • Degree level
    Master's
  • Department
    • Department of Electrical and Computer Engineering
  • Specialization
    • Digital Signals and Image Processing
  • Supervisor / co-supervisor and their department(s)
    • Zhao, Hong (Electrical and Computer Engineering)
    • Jiang, Hai (Electrical and Computer Engineering)
  • Examining committee members and their departments
    • Kong, Linglong (Mathematical and Statistical Sciences)
    • Niu,Di(Electrical and Computer Engineering)