本文提交几个有意思的点
（1）矩阵分解后，利用近邻提供对结果的解释，（如隐空间近邻喜欢了该物品，所以推荐给active user）。
（2）本文认为（1）中的缺点是，分解模型并不能保证：现实中相近的用户，在隐空间中，距离就一定近，因为objective function并没有保证这一点
本文核心句：（让跟用户相关的item在隐空间中的距离也近可能的近，近了就表明分解后的矩阵可以解释了？为什么不把缩短user和item之间距离改成缩短user和user之间的距离，不是更好的解决以上（2）的问题？）The Explainability term used in the objective function J , encourages items, that have higher explainability relative to a user, to be projected close to that user in the latent space, while keeping the rating prediction error small.
另外，本博文附带的另一篇相关文章：Explainable Matrix Factorization for Collaborative Filtering的核心思想是：直接建立隐空间用户和物品之间的距离联系。物品和item可以相关表示（解释）？具体做法是：在基本的矩阵分解objective function中增加一项约束，即目标用户（active user）同相似的物品的欧式距离近可能近(Ui-Vj)^2。其中，如何定义active user的相似用户：active user的相似users集合中，被访问越多的物品离active user越近（访问的用户数的比例可作为此约束的权值），论文中的objective function表述的很清楚。
| 文献题目 | 去谷歌学术搜索 | ||||||||||
| Using Explainability for Constrained Matrix Factorization | |||||||||||
| 文献作者 | Behnoush Abdollahi | ||||||||||
| 文献发表年限 | 2017 | ||||||||||
| 文献关键字 | |||||||||||
| factorization model; explainability; 根据近邻进行解释 | |||||||||||
| 摘要描述 | |||||||||||
| Accurate model-based Collaborative Filtering (CF) approaches, such as Matrix Factorization (MF), tend to be black-box machine learning models that lack interpretability and do not provide a straightforward explanation for their outputs. Yet explanations have been shown to improve the transparency of a recommender system by justifying recommendations, and this in turn can enhance the user’s trust in the recommendations. Hence, one main challenge in designing a recommender system is mitigating the trade-off between an explainable technique with moderate prediction accuracy and a more accurate technique with no explainable recommendations. In this paper, we focus on factorization models and further assume the absence of any additional data source, such as item content or user attributes. We propose an explainability constrained MF technique that computes the top-n recommendation list from items that are explainable. Experimental results show that our method is effective in generating accurate and explainable recommendations. | |||||||||||
