Abstract:

The major objective of recommendation algorithms is to accurately predict the rating value of items.However, it has been recognized that accurate prediction of rating values is not the only requirement for achieving users’ satisfaction. The diversity of recommendation lists has gained importance recently. In this paper, we propose a novel recommendation algorithm which focuses on longtail item promotion and aggregate diversity enhancement. We take predicted item values, item popularity, item preference into account. Experimental results show that this approach can discover more worthrecommending long tails items, and improve aggregate diversity while maintaining a certain level of accuracy.

Key words: recommendation system;diversity;coverage