Challenges of data privacy in recommender systems - Review of literature

No Thumbnail Available
Journal Title
Journal ISSN
Volume Title
School of Business | Bachelor's thesis
Electronic archive copy is available locally at the Harald Herlin Learning Centre. The staff of Aalto University has access to the electronic bachelor's theses by logging into Aaltodoc with their personal Aalto user ID. Read more about the availability of the bachelor's theses.
Degree programme
Tieto- ja palvelujohtaminen
Recommender systems are applications that are used in e-commerce platforms to personalize the content for each user in order to make the customer experience unique for each customer. This thesis aims to create a framework of a different kind of recommender systems the strengths and weaknesses of these systems, and discover if there are data privacy issues related to these systems and what are the methods that can be utilized in order to prevent this risks to capitalize. Recommender systems can be divided into two major groups: content-based (CB) and collaboration filtering (CF) systems. Pure CB or CF systems are rare as they both can be seen to have major weaknesses if used on their own, due to that most used recommendation systems now days are hybrid recommendation systems that use features from both CB and CF systems in order to lessen the weak points of each of these basic systems and gaining the benefits of both systems. Recommender systems are facing major data privacy issues, that is due to the fact that all the systems that personalize content according to the user need to have some amount of data from the user in order to make recommendations. These data privacy issues can be seen to be following: possible data breaches, recommendations revealing customers interests and the possibility of companies selling the data of customer’s preferences. There are privacy-enhancing techniques that can be utilized in order to lessen the chance of these data privacy issues causing major damage. This literature review recognizes four different categories on how to approach these threats: encryption-based methods, trusted third party methods, collaborative techniques and data-perturbation techniques. Privacy-enhancing techniques are often seen being a trade-off between accuracy of the recommendations and privacy or efficiency of the system and privacy. Because of these trade-offs there are currently no clear privacy-enhancing technique that could be said to work substantially better than others in every situation. It is clear that more research is needed on the subject of data privacy in recommendation systems to better understand these issues.
Thesis advisor
Kauppi, Katri
recommendation systems, data privacy, collaboration filtering, content-based recommendation
Other note