Adaptive Sequential Recommendation Using Context Trees

Machine learning is often used to acquire knowledge in domains that undergo frequent changes, such as networks, social media, or markets. These frequent changes pose a challenge to most machine learning methods as they have difficulty adapting. So my thesis topic focus on adaptive machine learning models. At the first step, we consider a forum content recommender system for massive open online courses (MOOCs) as an example of an application where recommendations have to adapt to new items and evolving user preferences. We formalize the recommendation problem as a sequence prediction problem and compare different recommendation methods, including a new method called context tree (CT). The results show that the CT recommender performs much better than other methods. We analyze the reasons for this and demonstrate that it is because of better adaptation to changes in the domain.