LibRec 精选：推荐的可解释性[综述]

推荐进展 第五期（更新至2018.5.3），分为社交关注和论文进展两部分，其中论文的摘要部分经过句子抽取算法处理过。

本期重点推荐论文部分的第一篇，即对推荐可解释性的综述文章，值得一读。另外，SIGIR 2018的录用论文里也有一些是做推荐的可解释性方面的工作。

1. SIGIR Accepted Papers: http://sigir.org/sigir2018/accepted-papers/，今年有很多的推荐系统方面的论文。

2. Mathematical Notation for Recommender Systems by Michael Ekstrand，这些也是我经常使用的数学符号。

论文进展

---

1. Explainable Recommendation: A Survey and New Perspectives

Yongfeng Zhang, Xu Chen

https://arxiv.org/abs/1804.11192v2

In recent years, a large number of explainable recommendation approaches -- especially model-based explainable recommendation algorithms -- have been proposed and adopted in real-world systems. 3) We summarize the application of explainable recommendation in different recommendation tasks, including product recommendation, social recommendation, POI recommendation, etc. We end the survey by discussing potential future research directions to promote the explainable recommendation research area.

2. A Missing Information Loss function for implicit feedback datasets

Juan Arévalo, Juan Ramón Duque, Marco Creatura

https://arxiv.org/abs/1805.00121v1

Missing information is often used as negative feedback. This is frequently done either through negative sampling (point-wise loss) or with a ranking loss function (pair- or list-wise estimation). In this paper we propose a novel objective function, the Missing Information Loss (MIL) function, that explicitly forbids treating unobserved user-item interactions as positive or negative feedback.

3. TR-SVD: Fast and Memory Efficient Method for Time Ranged Singular Value Decomposition

Jun-Gi Jang, Dongjin Choi, Jinhong Jung, U Kang

https://arxiv.org/abs/1805.00754v1

Singular value decomposition (SVD) is a crucial tool to discover hidden factors in multiple time series data, and has been used in many data mining applications including dimensionality reduction, principal component analysis, recommender systems, etc. In this paper, we propose TR-SVD (Time Ranged Singular Value Decomposition), a fast and memory efficient method for finding latent factors of time series data in an arbitrary time range. TR-SVD incrementally compresses multiple time series data block by block to reduce the space cost in storage phase, and efficiently computes singular value decomposition (SVD) for a given time range query in query phase by carefully stitching stored SVD results.

4. Propagation of content similarity through a collaborative network for live show recommendation

Jean Creusefond, Matthieu Latapy

https://arxiv.org/abs/1804.09073v1

We combine collaborative and content-based filtering to take benefit of past activity of users and of the features of the new show. Indeed, as this show is new we cannot rely on collaborative filtering only. To solve this cold-start problem, we perform network alignment and insert the new show in a way consistent with collaborative filtering.

5. A multi-level collaborative filtering method that improves recommendations

Nikolaos Polatidis, Christos K. Georgiadis

https://arxiv.org/abs/1804.08891v1

Collaborative filtering is one of the most used approaches for providing recommendations in various online environments. In this paper we propose a multi-level recommendation method with its main purpose being to assist users in decision making by providing recommendations of better quality. The proposed method can be applied in different online domains that use collaborative recommender systems, thus improving the overall user experience.