Improving the Performance of Recommendation Systems

The goal of this presentation is to show novel techniques that help to improve the accuracy as well as prediction latency of recommender systems. To improve the accuracy we propose new aggregated forms of user-item matrix by using clustering technique and utilizing 
contextual information. On the other hand, we combine factorization machines with association rules to improve the recommendation speed.

All presented methods and results in the presentation are either published or freshly submitted as a research paper.

However, this is also invitation for anyone who is interested in research collaboration in recommendation systems to contact with me for future cooperation.