Recommendations as a Game: Reinforcement Learning for Recommendation Engines - Realtime Data Mining - page 55

4.5 Summary
This chapter was devoted to the application of reinforcement learning to recom-
mendation engines. We have introduced RE-specific empirical assumptions to
reduce the complexity of RL in order to make it applicable to real-life recommen-
dation problems. Especially, we provided a new approach for estimating transition
probabilities of multiple recommendations from that of single recommendations.
Nevertheless, the estimation of transition probabilities for single recommendations
was left as an open problem that will be addressed in the next chapter. Finally, we
introduced a simple framework for testing online recommendations.