Intelligent User Preference Detection for Product Brokering

Abstract

We present a generic approach to capture individual user responding towards product attributes including non-quantifiable ones. The proposed solution does not generalize or stereotype user preference, but captures the user’s unique taste and recommends a list of products to the user. Under the proposed generic approach, the system is able to handle the inclusion of any unaccounted attribute that is not predefined in the system, without reprogramming the system. The system is able to cater for any unaccounted attribute through a general description field found in most product databases. This is extremely useful as hundreds of new attributes of products emerge each day, making any complex analysis impossible. In addition, the system is self-adjusting in nature and can adapt to changes in a user’s preference.