DELF: A Dual-Embedding based Deep Latent Factor Model for Recommendation

Cheng, Weiyu; Shen, Yanyan; Zhu, Yanmin; Huang, Linpeng

doi:10.24963/ijcai.2018/462

Cited by 48 publications

(53 citation statements)

References 7 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…From the figures, we have the following observations. First, DELF, ConvMF and CMN outperform NeuMF across different datasets and metrics, which is consistent with previous works [3,7,9]. The reason is that NeuMF mainly relies user-item interactions for exploiting user preferences, and it suffers from data sparseness due to the sparsity nature of the rating matrix.…”

Section: Recommendation Comparisonssupporting

confidence: 87%

“…ConvMF [9] proposes to use outer product to transform embeddings of each user-item pair into a two-dimensional interaction map, and then employs convolutional and pooling layer to model high-order interrelations among embedding dimensions. Cheng et al [3] leverage item contexts (i.e. historical items rated by users) to compensate the interaction function.…”

Section: Neural Recommendationmentioning

confidence: 99%

“…ConvNCF [9] utilizes outer product to transform each user-item embedding pair into a two-dimensional interaction map, and employs convolution and pooling layers for modeling high-order user-item correlations. [3] involves context items (e.g. items rated by users) for modeling user-item interactions, hence user preferences over items are comprised of user-item and item-item similarities.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

DBRec

Wen

Zhong

et al. 2019

Proceedings of the 28th ACM International Conference on Information and Knowledge Management

View full text Add to dashboard Cite

In recommender systems, the user-item interaction data is usually sparse and not sufficient for learning comprehensive user/item representations for recommendation. To address this problem, we propose a novel dual-bridging recommendation model (DBRec). DBRec performs latent user/item group discovery simultaneously with collaborative filtering, and interacts group information with users/items for bridging similar users/items. Therefore, a user's preference over an unobserved item, in DBRec, can be bridged by the users within the same group who have rated the item, or the user-rated items that share the same group with the unobserved item. In addition, we propose to jointly learn user-user group (itemitem group) hierarchies, so that we can effectively discover latent groups and learn compact user/item representations. We jointly integrate collaborative filtering, latent group discovering and hierarchical modelling into a unified framework, so that all the model parameters can be learned toward the optimization of the objective function. We validate the effectiveness of the proposed model with two real datasets, and demonstrate its advantage over the state-ofthe-art recommendation models with extensive experiments.

show abstract

Section: Recommendation Comparisonssupporting

confidence: 87%

Section: Neural Recommendationmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

DBRec

Wen

Zhong

et al. 2019

Proceedings of the 28th ACM International Conference on Information and Knowledge Management

View full text Add to dashboard Cite

show abstract

“…item) is a one-hot vector with high dimension, and the embedding operation is to project each user to a low-dimensional representation. Inspired by [15] and [2], each user can be represented by a user-speci c embedding and the items rated by her. e former representation re ects user's interests while the la er one (called item-based user embedding) captures implicit in uence of her rated history on current decision.…”

Section: Model Frameworkmentioning

confidence: 99%

Dual Graph Attention Networks for Deep Latent Representation of Multifaceted Social Effects in Recommender Systems

Zhang

Gao

et al. 2019

The World Wide Web Conference

279

152

View full text Add to dashboard Cite

Social recommendation leverages social information to solve data sparsity and cold-start problems in traditional collaborative ltering methods. However, most existing models assume that social e ects from friend users are static and under the forms of constant weights or xed constraints. To relax this strong assumption, in this paper, we propose dual graph a ention networks to collaboratively learn representations for two-fold social e ects, where one is modeled by a user-speci c a ention weight and the other is modeled by a dynamic and context-aware a ention weight. We also extend the social e ects in user domain to item domain, so that information from related items can be leveraged to further alleviate the data sparsity problem. Furthermore, considering that di erent social e ects in two domains could interact with each other and jointly inuence users' preferences for items, we propose a new policy-based fusion strategy based on contextual multi-armed bandit to weigh interactions of various social e ects. Experiments on one benchmark dataset and a commercial dataset verify the e cacy of the key components in our model. e results show that our model achieves great improvement for recommendation accuracy compared with other state-of-the-art social recommendation methods.

show abstract

“…3.1.2 high-order Information Aggregation. In collaborative filtering (CF) models [6,12,18] and sequential recommendation models [14,32,43], we only use the user (item) directly interacted items (users) to represent a user (item) which may cause a narrow understanding of user or item properties.…”

Section: Cross Neighbor Co-attention Networkmentioning

confidence: 99%

Sequential Recommendation with Dual Side Neighbor-based Collaborative Relation Modeling

Qin

Ren

et al. 2020

Proceedings of the 13th International Conference on Web Search and Data Mining

View full text Add to dashboard Cite

Sequential recommendation task aims to predict user preference over items in the future given user historical behaviors. The order of user behaviors implies that there are resourceful sequential patterns embedded in the behavior history which reveal the underlying dynamics of user interests. Various sequential recommendation methods are proposed to model the dynamic user behaviors. However, most of the models only consider the user's own behaviors and dynamics, while ignoring the collaborative relations among users and items, i.e., similar tastes of users or analogous properties of items. Without modeling collaborative relations, those methods suffer from the lack of recommendation diversity and thus may have worse performance. Worse still, most existing methods only consider the user-side sequence and ignore the temporal dynamics on the item side. To tackle the problems of the current sequential recommendation models, we propose Sequential Collaborative Recommender (SCoRe) which effectively mines high-order collaborative information using cross-neighbor relation modeling and, additionally utilizes both user-side and item-side historical sequences to better capture user and item dynamics. Experiments on three real-world yet large-scale datasets demonstrate the superiority of the proposed model over strong baselines.

show abstract

DELF: A Dual-Embedding based Deep Latent Factor Model for Recommendation

Cited by 48 publications

References 7 publications

DBRec

DBRec

Dual Graph Attention Networks for Deep Latent Representation of Multifaceted Social Effects in Recommender Systems

Sequential Recommendation with Dual Side Neighbor-based Collaborative Relation Modeling

Contact Info

Product

Resources

About