Collaborative filtering techniques have been studied extensively during the last decade. Many open source packages (Apache Mahout, LensKit, MyMediaLite, rrecsys etc.) have implemented them, but typically the top-N recommendation lists are only based on a highest predicted ratings approach. However, exploiting frequencies in the user/item neighborhood for the formation of the top-N recommendation lists has been shown to provide superior accuracy results in offline simulations. In addition, most open source packages use a time-independent evaluation protocol to test the quality of recommendations, which may result to misleading conclusions since it cannot simulate well the real-life systems, which are strongly related to the time dimension. In this paper, we have therefore implemented the time-aware evaluation protocol to the open source recommendation package for the R language — denoted rrecsys — and compare its performance across open source packages for reasons of replicability. Our experimental results clearly demonstrate that using the most frequent items in neighborhood approach significantly outperforms the highest predicted rating approach on three public datasets. Moreover, the time-aware evaluation protocol has been shown to be more adequate for capturing the life-time effectiveness of recommender systems.

中文翻译：

---

改进开源包中的时间感知建议

在过去的十年中，协同过滤技术得到了广泛的研究。许多开源软件包（Apache Mahout、LensKit、MyMediaLite、rrecsys 等）已经实现了它们，但通常前 N 推荐列表仅基于最高预测评级方法。然而，利用用户/项目邻域中的频率来形成前 N 个推荐列表已被证明可以在离线模拟中提供卓越的准确度结果。此外，大多数开源包使用与时间无关的评估协议来测试推荐的质量，这可能会导致误导性结论，因为它不能很好地模拟与时间维度密切相关的现实系统。在本文中，因此，我们为 R 语言的开源推荐包（表示为 rrecsys）实现了时间感知评估协议，并出于可复制性的原因比较了其跨开源包的性能。我们的实验结果清楚地表明，在三个公共数据集上，使用邻域中最常见项目的方法明显优于最高预测评级方法。此外，时间感知评估协议已被证明更适合捕获推荐系统的生命周期有效性。我们的实验结果清楚地表明，在三个公共数据集上，使用邻域中最常见项目的方法明显优于最高预测评级方法。此外，时间感知评估协议已被证明更适合捕获推荐系统的生命周期有效性。我们的实验结果清楚地表明，在三个公共数据集上，使用邻域中最常见项目的方法明显优于最高预测评级方法。此外，时间感知评估协议已被证明更适合捕获推荐系统的生命周期有效性。