Group Recommendation via Self-Attention and Collaborative Metric Learning Model

Wang, Haiyan; Li, Yuliang; Frimpong, Felix

doi:10.1109/access.2019.2953176

Cited by 14 publications

(6 citation statements)

References 26 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…SACML [43]: The Self-Attention and Collaborative Metric Learning model employs the self-attention mechanism to automatically learn the weight of each group member, which are then aggregated to form a group. After that, the collaborative metric learning technique is leveraged to obtain the group and item representations in the embedding space.…”

Section: Comparison Methodsmentioning

confidence: 99%

Overcoming Data Sparsity in Group Recommendation

Yin¹,

Wang²,

Zheng³

et al. 2020

IEEE Trans. Knowl. Data Eng.

View full text Add to dashboard Cite

It has been an important task for recommender systems to suggest satisfying activities to a group of users in people's daily social life. The major challenge in this task is how to aggregate personal preferences of group members to infer the decision of a group. Conventional group recommendation methods applied a predefined strategy for preference aggregation. However, these static strategies are too simple to model the real and complex process of group decision-making, especially for occasional groups which are formed ad-hoc. Moreover, group members should have non-uniform influences or weights in a group, and the weight of a user can be varied in different groups. Therefore, an ideal group recommender system should be able to accurately learn not only users' personal preferences but also the preference aggregation strategy from data. In this paper, we propose a novel end-to-end group recommender system named CAGR (short for "Centrality-Aware Group Recommender"), which takes Bipartite Graph Embedding Model (BGEM), the self-attention mechanism and Graph Convolutional Networks (GCNs) as basic building blocks to learn group and user representations in a unified way. Specifically, we first extend BGEM to model group-item interactions, and then in order to overcome the limitation and sparsity of the interaction data generated by occasional groups, we propose a self-attentive mechanism to represent groups based on the group members. In addition, to overcome the sparsity issue of user-item interaction data, we leverage the user social networks to enhance user representation learning, obtaining centrality-aware user representations. To further alleviate the group data sparsity problem, we propose two model optimization approaches to seamlessly integrate the user representations learning process. We create three large-scale benchmark datasets and conduct extensive experiments on them. The experimental results show the superiority of our proposed CAGR by comparing it with state-of-the-art group recommender models.

show abstract

Section: Comparison Methodsmentioning

confidence: 99%

Overcoming Data Sparsity in Group Recommendation

Yin¹,

Wang²,

Zheng³

et al. 2020

IEEE Trans. Knowl. Data Eng.

View full text Add to dashboard Cite

show abstract

“…However, a key weakness is their assumption that users have the same likelihood to follow individual and collective preferences, across different groups. Neural methods explore attention mechanisms [2] to learn data-driven preference aggregators [7,34,35]. MoSAN [34] models group interactions via sub-attention networks; however, MoSAN operates on persistent groups while ignoring users' personal activities.…”

Section: Related Workmentioning

confidence: 99%

“…where α u indicates the contribution of a user u towards the group decision. This can be trivially extended to item-conditioned weighting [7], self-attention [35] and sub-attention networks [34].…”

Section: 42mentioning

confidence: 99%

“…They mainly fall into two categories: heuristic pre-defined aggregation (e.g., least misery [3]) that disregards group interactions; data-driven strategies such as probabilistic models [27,40] and neural preference aggregators [7,34]. A key weakness is that these methods either ignore individual user activities [34,35] or assume that users have the same likelihood to follow individual and collective preferences, across different groups [7,27,40]. Lack of expressivity to distinguish the role of individual preferences across groups results in degenerate solutions for sparse ephemeral groups.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

GroupIM: A Mutual Information Maximization Framework for Neural Group Recommendation

Sankar,

Wu,

et al. 2020

Preprint

View full text Add to dashboard Cite

We study the problem of making item recommendations to ephemeral groups, which comprise users with limited or no historical activities together. Existing studies target persistent groups with substantial activity history, while ephemeral groups lack historical interactions. To overcome group interaction sparsity, we propose data-driven regularization strategies to exploit both the preference covariance amongst users who are in the same group, as well as the contextual relevance of users' individual preferences to each group.We make two contributions. First, we present a recommender architecture-agnostic framework GroupIM that can integrate arbitrary neural preference encoders and aggregators for ephemeral group recommendation. Second, we regularize the user-group latent space to overcome group interaction sparsity by: maximizing mutual information between representations of groups and group members; and dynamically prioritizing the preferences of highly informative members through contextual preference weighting. Our experimental results on several real-world datasets indicate significant performance improvements (31-62% relative NDCG@20) over state-of-the-art group recommendation techniques. CCS CONCEPTS• Information systems → Recommender systems; • Computing methodologies → Neural networks.

show abstract

“…In the past few decades, personalized recommendation technology has gradually expanded from movies (Halder et al , 2012), music (Jun et al , 2010) and hotels (Chang et al , 2013) to short videos (Wang et al , 2019b; Li et al , 2019) and social networking sites (Tang et al , 2016; Yu and Li, 2018) because of the continuous growth of Web data and the rapid growth of various new media industries (Shao et al , 2021). The growth of data and the change in applications have also made researchers gradually realize the importance of studying group activities, thus giving rise to the shift in personalized recommendation service targets from individuals to groups (Huang et al , 2020).…”

Section: Introductionmentioning

confidence: 99%

A group recommender system for books based on fine-grained classification of comments

Xiong

Guo

et al. 2023

View full text Add to dashboard Cite

Purpose The purpose of this study is to investigate how book group recommendations can be used as a meaningful way to suggest suitable books to users, given the increasing number of individuals engaging in sharing and discussing books on the web. Design/methodology/approach The authors propose reviews fine-grained classification (CFGC) and its related models such as CFGC1 for book group recommendation. These models can categorize reviews successively by function and role. Constructing the BERT-BiLSTM model to classify the reviews by function. The frequency characteristics of the reviews are mined by word frequency analysis, and the relationship between reviews and total book score is mined by correlation analysis. Then, the reviews are classified into three roles: celebrity, general and passerby. Finally, the authors can form user groups, mine group features and combine group features with book fine-grained ratings to make book group recommendations. Findings Overall, the best recommendations are made by Synopsis comments, with the accuracy, recall, F-value and Hellinger distance of 52.9%, 60.0%, 56.3% and 0.163, respectively. The F1 index of the recommendations based on the author and the writing comments is improved by 2.5% and 0.4%, respectively, compared to the Synopsis comments. Originality/value Previous studies on book recommendation often recommend relevant books for users by mining the similarity between books, so the set of book recommendations recommended to users, especially to groups, always focuses on the few types. The proposed method can effectively ensure the diversity of recommendations by mining users’ tendency to different review attributes of books and recommending books for the groups. In addition, this study also investigates which types of reviews should be used to make book recommendations when targeting groups with specific tendencies.

show abstract

Group Recommendation via Self-Attention and Collaborative Metric Learning Model

Cited by 14 publications

References 26 publications

Overcoming Data Sparsity in Group Recommendation

Overcoming Data Sparsity in Group Recommendation

GroupIM: A Mutual Information Maximization Framework for Neural Group Recommendation

A group recommender system for books based on fine-grained classification of comments

Contact Info

Product

Resources

About