Description

Have you ever been amazed that the products an online retailer suggested to you are just what you want to buy? Do you want to know what tricks are played in the background for such intelligent recommendations? These intelligent recommendations are owed to the recommender system, which plays an important role in most online retailers and social networking services, such as Netflix and Amazon. A good recommender system should be clever enough to learn users’ preferences from their historical records (e.g., ratings, clicks, and reviews) and makes personalized recommendations in future. Nowadays, the mainstream technology for personalized recommendation is collaborative filtering (CF). Collaborative filtering is a computational approach, which predicts a user’s preference by finding like-minded users based on their historical records. In the last two decades, a large number of CF algorithms have been developed as the state-of-the-art techniques underlying most commercial recommender systems. To introduce this rising technology to broad audience in KAUST, we provide a short course including three sessions: (1) An overview of real-world recommender systems; (2) State-of-the-art CF techniques; and (3) Challenges in CF-based recommender systems.