Large 00238

Product Recommenders for Online Users

Pearl Pu

Recorded 05 May 2011 in Lausanne, Vaud, Switzerland

Event: KTN - Know Thy Neighbor

Abstract

My research is multi-disciplinary and focuses on issues in the intersection of human computer interaction, artificial intelligence, and behavioral decision theories. In my earlier work, I considered the interaction of artificial intelligence and visualization techniques in various decision support systems, for example for aircraft allocation and network configuration.

In the last few years I have focused my attention on designing product recommenders in online environments. First, I present product recommenders as a special case of multi-criteria utility problems: how to help a user find the most preferred item in a large set of options. Then I focus on the main user challenge in such recommender systems, which is users' expectation of high decision quality while unwilling or unable to accurately state their preferences. To be able to solve this problem is crucial in motivating users to accept items recommended to them, especially in e-commerce environments. I will describe the user task analysis process that has guided us in coming up with an innovative solution, called Example Critiquing. I will highlight the design features of Example Critiquing and show how it is sensible to users' behaviors and needs. I will also present some major outcomes of validating the system with real users.

Through the presentation of this research work, I hope to demonstrate the results of combining user experience research from HCI and intelligent techniques from AI to achieve interactive intelligence: how humans and machines can combine their abilities to solve difficult problems.

Watched 487 times.

 Watch