automated website synthesis and uncertainty reasoning for customer preference management

Siu-wai Leung
University of Edinburgh
Siu-ki Leung
Mobile Data Solutions
Andy Kung
The Prudential Assurance Company
Tommy Tang
ACORN (Market Research)
Kai-Fai Wong
Independent Statistics Consultant
Tony Li
Phillip Morris Asia


Reasoning and acting on the basis of customer preference are necessary for customer relationship management (CRM) systems to provide the most preferred products or services at the right time to every customer (Leung et al, 2002b). The holy grail of developing web-based CRM systems or software agents is designing scientific and efficient algorithms that can automatically 1) reason about customer preference under uncertainty and 2) synthesize a customized web interface. We refer to such CRM systems and software agents as CRM agents hereafter. Due to resource constraints and technical limitations in real world, exhaustive knowledge of customer preference cannot be made available to CRM agents. The capability of CRM agents to reason about customer preference under uncertainty is crucial for proper actions. When interacting with different customers through the web, the form and content of web should be reconfigured according to individual customer preference, i.e., personalization. Automated techniques in website synthesis are greatly desirable because of the high cost and inefficiency of manual web crafting. As shown in figure 1, all these are key functions and processes of customer preference management (CPM). In the following sections, we describe about preference elicitation, uncertainty handling in preference reasoning, and automated website synthesis based on customer preference.

Preference Elicitation