Amy B.Z. Zhang
Personalized Recommender Systems: Technological Impact and Concerns
Most of our online activities are, at least in part, powered by personalized recommender systems. While automatic pattern extraction as a technology holds great promise, it can also have alarming adverse impacts. This talk will give a high-level overview of common techniques for personalized recommender systems, and how they connect to problems on both the personal and social level. It will also discuss some alternative approaches addressing these issues, and why a solution cannot come from technology alone.
Amy Zhang is a PhD Candidate in Operations Research and Information Engineering (ORIE) at Cornell Tech, working with Prof. Itai Gurvich on approximate control optimization in an evolving system. Her research focuses on approximating large systems by a small number of clusters in settings that can be modeled as Markov Decision Processes (MDP). She would like to extend the results to applications where “best” actions are learned through interactions, in particular personalized recommendations, and enquire into what qualifies as “best”: 1. How to tradeoff/ combine the profit goals of the entity making the recommendation vs the value to the person receiving the recommendation? How to account for the human part of the value such as feelings in algorithmic design? 2. Could the clustering structure in the approximation be leveraged to reduce the “echo chamber” effect in certain contexts such as media or news? Or going further, how much should an algorithm be concerned about better vs preferred? 3. Might there be potential implications of the technique on the ability to replace individual data with group level ones? Is it still valuable if an algorithm is not privacy preserving but makes the user feel less “watched”? As a DLI Doctoral Fellow, Amy seeks to pursue discussions underlying these questions and explore what downstream societal impacts can be contained at the level of algorithm design.