Kartik Sharma,Yeon-Chang Lee,Sivagami Nambi,Aditya Salian,Shlok Shah,Sang-Wook Kim,Srijan Kumar
Abstract:Social recommender systems (SocialRS) simultaneously leverage the user-to-item interactions as well as the user-to-user social relations for the task of generating item recommendations to users. Additionally exploiting social relations is clearly effective in understanding users’ tastes due to the effects of homophily and social influence. For this reason, SocialRS has increasingly attracted attention. In particular, with the advance of graph neural networks (GNN), many GNN-based SocialRS methods have been developed recently. Therefore, we conduct a comprehensive and systematic review of the literature on GNN-based SocialRS. In this survey, we first identify 84 papers on GNN-based SocialRS after annotating 2,151 papers by following the PRISMA framework (preferred reporting items for systematic reviews and meta-analyses). Then, we comprehensively review them in terms of their inputs and architectures to propose a novel taxonomy: (1) input taxonomy includes 5 groups of input type notations and 7 groups of input representation notations; (2) architecture taxonomy includes 8 groups of GNN encoder notations, 2 groups of decoder notations, and 12 groups of loss function notations. We classify the GNN-based SocialRS methods into several categories as per the taxonomy and describe their details. Furthermore, we summarize benchmark datasets and metrics widely used to evaluate the GNN-based SocialRS methods. Finally, we conclude this survey by presenting some future research directions. GitHub repository with the curated list of papers are available at https://github.com/claws-lab/awesome-GNN-social-recsys.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is how to effectively use Graph Neural Networks (GNN) in Social Recommendation Systems (SocialRS) to improve the performance of the recommendation system. Specifically, the paper focuses on how to generate more accurate item recommendations by combining user - item interactions and user - user social relations. The paper points out that using social relations can lead to a better understanding of users' preferences, because these relations are affected by the effects of homophily and social influence. Therefore, the paper aims to conduct a comprehensive and systematic literature review of GNN - based social recommendation systems to identify and analyze existing methods, challenges, and future research directions.
The main contributions of the paper include:
1. **The first systematic review**: As far as the authors know, this is the first systematic review specifically for GNN - based social recommendation system methods. Most existing reviews mainly focus on traditional methods (such as matrix factorization), feature information, or specific applications, while ignoring methods using modern deep - learning techniques, especially GNN.
2. **Comprehensive review**: The researchers systematically identified relevant literature following the PRISMA framework and comprehensively reviewed 84 papers on GNN - based social recommendation systems in terms of input and architecture. The paper also provides a timeline (see Figure 2) showing the development process of GNN - based social recommendation system methods.
3. **Novel taxonomy**: A new taxonomy of GNN - based social recommendation system inputs and architectures is proposed, including 5 groups of input type representations, 7 groups of input representation representations, 8 groups of GNN encoder representations, 2 groups of decoder representations, and 12 groups of loss function representations (of which 4 are main losses and 8 are auxiliary losses).
4. **Benchmark datasets**: 17 benchmark datasets used to evaluate the performance of GNN - based social recommendation systems are reviewed and divided into 8 domains (products, locations, movies, images, music, bookmarks, Weibo, and others). In addition, some statistical information for each dataset and a list of papers using the dataset are also provided.
5. **Future research directions**: The limitations of existing GNN - based social recommendation system methods are discussed, and several future research directions are proposed.
Through the above contributions, the paper provides researchers with a comprehensive resource for understanding the field of GNN - based social recommendation systems, which is helpful for promoting further research and development in this field.