Paper 6
A Framework for Modeling, Computing and Presenting Time-aware RecommendationsAuthors: Kostas Stefanidis, Eirini Ntoutsi, Mihalis Petropoulos, Kjetil Nørvåg, Hans-Peter Kriegel |
AbstractLately, recommendation systems have received significant at- tention. Most existing approaches though, recommend items of potential interest to users by completely ignoring the temporal aspects of rat- ings. In this paper, we argue that time-aware recommendations need to be pushed in the foreground. We introduce an extensive model for time-aware recommendations from two perspectives. From a fresh-based perspective, we propose using different aging schemes for decreasing the effect of historical ratings and increasing the infuence of fresh and novel ratings. From a context-based perspective, we focus on providing dif- ferent suggestions under different temporal specifications. To facilitate user browsing, we propose an effective presentation layer for time-aware recommendations based on user preferences and summaries for the sug- gested items. Our experiments with real movies ratings show that time plays an important role in the recommendation process. |
|