Thesis Proposal Page for Aniko Hannak

Proposal

• Personalization in Online Services: Measurement, Analysis, and Implications

Abstract

• An increasing number of web services are now using personalization algorithms to shape the content they serve to their users in order to best meet users’ tastes and needs. In many cases, personalization provides advantages for users: for example, users now expect search engines to return local results when searching for restaurants. However, the increasing level of personalization and the lack of transparency is now leading to concerns about its potential negative effects: in case of web search engines, personalization might lead to users not being able to access information that the search engine’s algorithm decides is irrelevant. Or, in case of e-commerce sites, it might lead to price discrimination, where different users see different prices or products when shopping online. Despite these concerns, there has been little quantification of the extent of personalization on Web services, or the user attributes that cause it.
  
  In this thesis proposal, I aim to develop the necessary tools to detect and quantify personalization. I first develop a general methodology for measuring personalization that can be used across multiple different web services. While conceptually simple, there are numerous details that the methodology must handle in order to accurately attribute differences in the content served to personalization. I collect data from real-world users to quantify the level of personalization observed today; I then build synthetic user accounts to investigate which user features drive personalization. This work is the first step towards understanding the extent and effects of personalization. Overall, the goal of this work is to both increase transparency into the operation of the web services as well as provide tools for the research community to further investigate the issue in other contexts.

Author

• Aniko Hannak

Committee

• Prof. David Lazer (Advisor)
• Sune Lehman (External Examiner, Technical University of Denmark)
• Prof. Alan Mislove (Advisor)
• Prof. Christo Wilson