Evaluating narrative-driven movie recommendations on Reddit
Evaluating narrative-driven movie recommendations on Reddit
复制标题
评估 Reddit 上叙事驱动的电影推荐
DOI：
10.1145/3301275.3302287
复制
发表时间：
2019
期刊：
影响因子：
--
通讯作者：
D. Helic
中科院分区：
文献类型：
--
作者：
Lukas Eberhard;Simon Walk;Lisa Posch;D. Helic
推荐系统已经成为无所不在的工具，被各种各样的用户用于日常生活任务，例如在网络商店或在线电影流媒体门户中查找产品。然而，在用户已经知道他们正在寻找什么的情况下（例如，“指环王”，但在空间与黑暗的氛围），大多数传统的推荐算法努力充分解决这样的先验定义的要求。因此，用户已经建立了专门的讨论板来向同行征求建议，这些建议理想地满足了所述要求。在本文中，我们开始确定的效用，完善的推荐算法计算建议时，提供这样的叙述。为此，我们首先从人类电影建议中众包参考评估数据集。我们使用这个数据集来评估五种推荐算法将这种叙述纳入其推荐的潜力。此外，我们为其他研究人员提供数据集，以推进叙事驱动推荐领域的研究现状。最后，我们使用我们的评估数据集来改进我们的算法推荐，以及IMDb现有的经验推荐。我们的研究结果表明，实现的推荐算法产生了非常不同的建议比人类提出相同的先验要求时。然而，通过精心配置的后过滤技术，我们可以超过基线高达100%。这代表了迈向更精细的算法叙事驱动推荐的重要第一步。
Recommender systems have become omni-present tools that are used by a wide variety of users in everyday life tasks, such as finding products in Web stores or online movie streaming portals. However, in situations where users already have an idea of what they are looking for (e.g., 'The Lord of the Rings', but in space with a dark vibe), most traditional recommender algorithms struggle to adequately address such a priori defined requirements. Therefore, users have built dedicated discussion boards to ask peers for suggestions, which ideally fulfill the stated requirements. In this paper, we set out to determine the utility of well-established recommender algorithms for calculating recommendations when provided with such a narrative. To that end, we first crowdsource a reference evaluation dataset from human movie suggestions. We use this dataset to evaluate the potential of five recommendation algorithms for incorporating such a narrative into their recommendations. Further, we make the dataset available for other researchers to advance the state of research in the field of narrative-driven recommendations. Finally, we use our evaluation dataset to improve not only our algorithmic recommendations, but also existing empirical recommendations of IMDb. Our findings suggest that the implemented recommender algorithms yield vastly different suggestions than humans when presented with the same a priori requirements. However, with carefully configured post-filtering techniques, we can outperform the baseline by up to 100%. This represents an important first step towards more refined algorithmic narrative-driven recommendations.