department of informatics

Recommender Systems FS-2016

18. February 2016 - 30. May 2016

The objectives of the course are:

  • To understand the basic concepts of RSs
  • Getting in touch with a number of RSs algorithms
  • To Learn about the different evaluation methods for RSs
  • Using a taxonomy, students will be able to classify different RSs solutions
  • At the end of the course, students will be able to develop a recommender system solution for a web shop

Recommender systems (RSs) are computer-based techniques that attempt to present information about products that are likely to be of interest to a user. These techniques are mainly used in Electronic Commerce (eCommerce) in order to provide suggestions on items that a customer is, presumably, going to like. Nevertheless, there are other applications that make use of RSs, such as social networks and community-building processes, among others. A recommender system is a specific type of information filtering technique that tries to present users with information about items (movies, music, books, news, web pages, among others) in which they are interested.

The term “item” is used to denote what the system recommends to users. To achieve this goal, the user profile is contrasted with the characteristics of the items. These features may come from the item content (content-based approach) or the user’s social environment (Collaborative Filtering). The use of these systems is becoming increasingly popular in the Internet because they are very useful to evaluate and filter the vast amount of information available on the Web in order to assist users in their search processes and retrieval.

RSs have been highly used and play an important role in different Internet sites that offer products and services in social networks, such as Amazon, YouTube, Netflix, Yahoo!, TripAdvisor, Facebook, and Twitter, among others. Many different companies are developing RSs techniques as an added value to the services they provide to their subscribers.


During the development of the course, the following topics will be presented:

  • Introduction to Recommender Systems. In this section, an brief introduction and history of RSs is presented. A classification of different approaches and techniques for RSs together with various examples from existing web shops will be shown.
  • Taxonomy of Recommender Systems. In the second part, a taxonomy used to classify different RSs solutions is presented. It helps students to understand the scope and objectives for the development of a RS.
  • Non-personalized Recommenders. In this section, the value of non-personalized RS, domains of usability as well as drawbacks of such applications is presented. Additionally, a review of aggregated opinion recommenders and basic product association recommenders is made.
  • Content-based Recommender Systems. In this section, students will be able to understand the problem that requires weighting for search or filtering. Additionally, a brief introduction to TFIDF weighting, and how it is used in both search and filtering. Finally, similarities and differences between content filtering and search will be presented.
  • User-User Recommender Systems. In this section, and intuition and history of the user-user collaborative filtering algorithm is presented. A review of the basic ideas and assumptions as well as limitations behind the algorithm is show.
  • Item-Item Recommender Systems. In this section, the so-called method Item-item collaborative filtering is presented. It is based on rating similarity between items.
  • Dimensionality Reduction for Recommenders Systems. In this section, the study of dimensionality reduction for recommender systems is presented. This methodology is mainly used to avoid performance complications related to large-scale datasets.
  • Evaluation Methods. At the end of this section, students will be able to understand the different ways of evaluating the goodness of a recommendation, recommender algorithm or system. The following metrics will be studied: accuracy, error, decision-support, user and usage-centered. Students will be able to understand how predictions and recommendations are evaluated.


  • The latest information and documents are available on Moodle. The password will be provided during the kick-off meeting. Students MUST be present in the kick-off meeting. Working groups will be organised. We set a limit of 10 groups (2 students/group). No exceptions will be made. 



Information about lecures, slides, and documents are available on Moodle

  • Lectures: mondays from 09h15-11h00. Room: B207
  • Exercise sessions: mondays from 14h15-15h00. Room: B230
  • Master of Science in Computer Science
  • Master of Arts in Information Management or Management and Business Communications


The course is evaluated as follows: 75% corresponds to student projects (more details below) and a final exam that represents the 25%.

Student Projects

Students will work in groups (2 students/group), each group will work on the development of a recommender system prototype, a research paper analysis, or a case study. Prototypes will be uploaded as part of the IS workbench:

The following steeps will be followed by groups in the course:

  • Project proposal: student groups will provide a proposal of project, which includes the following evaluation criteria: objectives (15%), background (15%), problem statement (15%), research questions (20%), methodology (30%), table of contents and References (5%). Proposal represents 10% of project final grade.
  • Midterm Appointment: student groups should make a short presentation (20 minutes) of the advances of the project. Presentations will be scheduled in the middle of the semester. Evaluation of presentations is based on the following criteria: presentation (40%), own contribution (25%), methodology (20%), thesis structure (10%), references (5%). Midterm appointment represents 10% of the project final grade.
  • Final Presentations: projects and prototypes will be presented in the final session, all students should present. It represents 5% of the project final grade.
  • Report Submission: Thesis reports are evaluated using the guidelines on how to write a scientific work from the IS group1. Report should be written in English with a maximum of 20-25 pages (content only). Index, references, etc. are not part of the report. Final report represents 75% of the project final grade.

Final Exam

The final exam represents 25% of the final grade and will include the theory and exercises that are part of the course.