Time-aware novelty metrics for recommender systems

Abstract

Time-aware recommender systems is an active research area where the temporal dimension is considered to improve the effectiveness of the recommendations. Even though performance evaluation is dominated by accuracy-related metrics – such as precision or NDCG –, other properties of the recommended items like their novelty and diversity have attracted attention in recent years, where several metrics have been defined with this goal in mind. However, it is unclear how suitable these metrics are to measure novelty or diversity in temporal contexts. In this paper, we propose a formulation to capture the time-aware novelty (or freshness) of the recommendation lists, according to different time models of the items. Hence, we provide a measure to account for how much a system is promoting fresh items in its recommendations. We show that time-aware recommenders tend to provide more fresh items, although this is not always the case, depending on statistical biases and patterns inherent to the data. Our results, nonetheless, indicate that the proposed formulation can be used to extend the knowledge about what items are being suggested by any recommendation technique aiming to exploit temporal contexts.