|
Department of
Computational
Perception
|
|
Spezielle Kapitel aus Informatik: RECOMMENDER SYSTEMS
344.033, KV, 2h, WS 2011/12
Univ.-Ass. Dr. Peter Knees
Time: Tuesday, 13:45 - 15:15
Location: HS 11
This lecture will be held in English.
Presentation slides will be made available prior to each class via KUSSS and serve as supplementary material for the course.
Description
Over the past decade, recommendation systems have become a ubiquitous
commodity on the Web. Popular Internet services and retailers like
Amazon.com, Last.fm, or Netflix provide their users with
recommendations of potentially interesting items and present
user-personalized views of their catalogues. As there is a potential
customer for even the most obscure product (i.e., products to be found
in the so-called "long tail"), methods to recommend appropriate items
at the right time are crucial for maximizing sales and user
satisfaction.
This lecture gives an introduction to current techniques for making
automatic recommendations. To this end, different methods to identify
items that could be relevant to a user (e.g., by analyzing usage data
of other users or by calculating the similarity of items) are presented
and discussed. Furthermore, important aspects like evaluation of
systems, their susceptibility to manipulation, and issues of the
presented methods are covered. Emphasis is given to two popular
application areas, namely music and movie recommendation.
Schedule (preliminary)
- 04. Oct. Overview of the Lecture, Introduction to Recommender Systems
- 11. Oct. Applications, "The Long Tail"
- 18. Oct. Knowledge-based Recs. vs. Collaborative Filtering
- 25. Oct. - No class! -
- 08. Nov. User-based Recs., User Modelling and Personalization, Relevance Feedback
- 15. Nov. Item-based Recs., Feature Extraction
- 22. Nov. Hybrid Recs. and Ensembles
- 29. Nov. Evaluating Recommender Systems
- 06. Dec. Issues: Data Sparsity, Cold Start, Serendipity
- 13. Dec. Hacking Systems & Users, Consumer Decision Making
- 10. Jan. Case-Study 1: Music Recommendation
- 17. Jan. Case-Study 2: Movie Recommendation, The Netflix Prize
- 24. Jan. Written Examination
Contact for Questions
Peter Knees
Tel. 2468-1513
last
edited by pk
at 2011-09-27