Hybrid Recommendation Base on Learning to Rank
Bin Xie, Xinhuai Tang, Feilong Tang · 2015
In order to solve the problem of recommender system using in different scenarios and the ranking of recommendation result, we propose a method using learning to rank for hybrid recommendation. It uses boosting merging algorithm as the base model, Lambda MART algorithm for updating. This makes our ranking model can be updated in real time based on user feedback information. By learning different data from different scenarios, the recommender system can be applied to different applications. In the end, we experiment our hybrid recommendation model by ranking evaluation NDCG.