Traditional recommender systems are based on Collaborative Filtering building models based on large user-item matrices. The weakness of these systems is that the adaptation to new data is resource-consuming. The algorithms fail in very dynamic domains where items continuously emerge and extend collections, and where existing items become less and less relevant. Examples include news, microblogs, or advertisement recommendations where content comes in the form of a constant stream of data.
For data stream-based scenarios characterized by rapidly changing sets of users and items the recommender algorithms must be adapted in order to be capable of the dynamics, ensuring that the recommender models keep track of the dynamic. This tutorial focused on real-time, stream-based recommenders tailored for providing high-quality recommendation results and ensuring technical constraints.
This tutorial introduces a combined view on multi-dimensional benchmarking and real-time requirements in recommender systems. Participants will not only learn about recommendation algorithms but also delve into practical issues related to system maintenance, model updating, and continuous evaluation. The tutorial offers access to a large-scale data stream for participants to experience the conditions faced by publishers with millions of readers. The tutorial supports participants to develop skills necessary for a successful career in computational intelligence.
- Slides [to be added]
- Templates [to be added]
- Data resources [to be added]
- References [to be added]