As a part of the Spring into STEM Lecture Series, UCL School of Management’s Yiting Deng will be hosting a session on; Applying Statistics to Improve Online Recommendations. In the talk, Yiting discusses the challenges in recommendation systems and why only a few (if any) address and accommodate these challenges.
Recommendation systems for contexts such as online news or online retailing are prevalent in our daily life. There are several challenges in such recommendation systems:
i) how to make initial recommendations to users with little or no response history
ii) how to learn user preferences for items
iii) how to scale the recommendation system across many users and items, with many potential demographics and attributes respectively. While many recommendation systems accommodate aspects of these challenges, few if any address all
We design a statistical method that can efficiently handle these issues simultaneously. Our online experiments in an online retail setting demonstrate the advantage of this method in comparison to existing methods.