Various Aspects of Recommender Systems

Recommender Systems

The goal of a Recommender System (RS) is to provide useful suggestions for a user and therefore to support her in a decision-making process. These systems are playing an increasingly important role. Many on-line retailers, streaming services and web applications such as Amazon.com, Spotify, Youtube, Netflix, Booking.com, Last.fm, etc. aim at improving the quality of their recommendations as a part of their offered services.

Many computer science fields are contributing to develop more sophisticated RSs. Information Retrieval (IR) share with RSs similar challenges, e.g. filtering and ranking. Therefore, IR techniques have been traditionally used to retrieve relevant items tailored for the users' needs. Data Mining has brought contributions in terms of dimensionality reduction, classification techniques (e.g. Bayesian networks or support vector machines), clustering techniques (e.g. k-means algorithm) and association rules. More recently, machine learning techniques are used to infer a model of user interests or to use these interests to categorize new items. The contribution is not limited to these fields: human-computer interaction, psychology and marketing are among the fields contributing to the multi-disciplinary RSs.

Master Project

The master projects have been designed to overcome a specific challenge of recommending items to users. The technique used will then depend on the kind of problem.

For a detail list of the offered projects click here.

Datasets*

These are some of the most commonly datasets used in our projects.

* DISCLAIMER: these datasets have been obtained or crawled by other academic institutions or were published by companies for research purposes. We can obtain permission restricted to academic usage. University of Freiburg neither owns intellectual property of the data nor is responsible for its content.

Curriculum

Master of Science: 3rd Semester (Teamproject / Masterproject)
ECTS: 16