Self-Supervised Hypergraph Convolutional Networks for Session-based Recommendation

Xia, Xin; Yin, Hongzhi; Yu, Junliang; Wang, Qinyong; Cui, Lizhen; Zhang, Xiangliang

doi:10.1609/aaai.v35i5.16578

Cited by 330 publications

(92 citation statements)

References 35 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“… DHCN [13]: It uses a hypergraph-based neural network to get higher-order information in items, and uses selfsupervised learning as an adjunct to enhanced hypergraph modeling.…”

Section: Comparison Methodsmentioning

confidence: 99%

See 1 more Smart Citation

Session recommendation based on target attention and directed graph

Sun²,

Liu³

et al. 2022

5th International Conference on Computer Information Science and Application Technology (CISAT 2022)

View full text Add to dashboard Cite

Session-based recommendations have aroused widespread concern, and they can predict the next item of interest in the chronological order according to a given anonymous behaviour sequence. Current session recommendations benefit from the ability of graph neural network to obtain the complicated transformation patterns between items, but they still have shortcomings. Most recommendation models based on graph neural network only consider the information between items in a session but not the target item information. This paper thus proposes a session recommendation model that combines the target attention mechanism and directed graph, which considers information between items in a session as well as information about the target item. The session-level embedding is obtained using an attention mechanism and a graph neural network, followed by the target embedding obtained using a target attention mechanism, and finally the session-level and target embeddings are stitched together to get the potential transformation information of items in the session. The model has experimented on two publicly available datasets. Compared with the existing benchmark model, the proposed model has improved in all indicators, proving the effectiveness of the proposed method.

show abstract

“… DHCN [13]: It uses a hypergraph-based neural network to get higher-order information in items, and uses selfsupervised learning as an adjunct to enhanced hypergraph modeling.…”

Section: Comparison Methodsmentioning

confidence: 99%

“…This approach yielded positive results. Xia et al [13] proposed a DHCN model to capture higher-order information between items through hypergraph-based neural networks. However, all these models ignore information about target items.…”

Section: Related Workmentioning

confidence: 99%

Session recommendation based on target attention and directed graph

Sun²,

Liu³

et al. 2022

5th International Conference on Computer Information Science and Application Technology (CISAT 2022)

View full text Add to dashboard Cite

show abstract

“…GCE-GNN [42] proposes to capture both global-level and sessionlevel interactions and aggregates item information through graph convolution and self-attention mechanism. Xia et al [45,46] proposed to integrate self-supervised learning into session-based recommendation to boost recommendation performance. These models are of great capacity in generating accurate recommendations but they were designed for server-side use, which cannot run on resource-constrained devices like smartphones.…”

Section: Session-based Recommendationmentioning

confidence: 99%

Efficient On-Device Session-Based Recommendation

Xia¹,

Yu²,

Wang³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

On-device session-based recommendation systems have been achieving increasing attention on account of the low energy/resource consumption and privacy protection while providing promising recommendation performance. To fit the powerful neural session-based recommendation models in resource-constrained mobile devices, tensor-train decomposition and its variants have been widely applied to reduce memory footprint by decomposing the embedding table into smaller tensors, showing great potential in compressing recommendation models. However, these model compression techniques significantly increase the local inference time due to the complex process of generating index lists and a series of tensor multiplications to form item embeddings, and the resultant on-device recommender fails to provide real-time response and recommendation. To improve the online recommendation efficiency, we propose to learn compositional encoding-based compact item representations. Specifically, each item is represented by a compositional code that consists of several codewords, and we learn embedding vectors to represent each codeword instead of each item. Then the composition of the codeword embedding vectors from different embedding matrices (i.e., codebooks) forms the item embedding. Since the size of codebooks can be extremely small, the recommender model is thus able to fit in resource-constrained devices and meanwhile can save the codebooks for fast local inference. Besides, to prevent the loss of model capacity caused by compression, we propose a bidirectional self-supervised knowledge distillation framework. Extensive experimental results on two benchmark datasets demonstrate that compared with existing methods, the proposed on-device recommender not only achieves an 8x inference speedup with a large compression ratio but also shows superior recommendation performance. The code is released at https://github.com/xiaxin1998/EODRec.

show abstract

“…In SBR task, Li et al [48] made use of a global-level contrastive learning model to solve noise and sampling problems in heterogeneous graphs. S 2 -DHCN [49] is the most relevant work to us, which designs a contrastive learning mechanism to enhance hyper-graph modeling via another line GCN model. But it still suffers from temporal information loss in the spatial structure, leading to sub-optimal performance.…”

Section: Contrastive Learning In Rsmentioning

confidence: 99%

“…• TASRec [14] incorporate temporal information via constructing a sequence of dynamic graph snapshots at different timestamps. • S 2 -DHCN [49] transforms the session data into hypergraph and line-graph and and uses self-supervised learning to enhance session-based recommendation.…”

Section: Baselinesmentioning

confidence: 99%

Spatio-Temporal Contrastive Learning Enhanced GNNs for Session-based Recommendation

Wan¹

2022

Preprint

View full text Add to dashboard Cite

<p>Session-based recommendation (SBR) systems aim to utilize the user’s short-term behavior sequence to predict the next item without the detailed user profile. Most recent works try to model the user preference by treating the sessions as between-item transition graphs and utilize various graph neural networks (GNNs) to encode the representations of pair-wise relations among items and their neighbors. Some of the existing GNN-based models mainly focus on aggregating information from the view of spatial graph structure, which ignores the temporal relations within neighbors of an item during message passing and the information loss results in a sub-optimal problem. Other works embrace this challenge by incorporating additional temporal information but lack sufficient interaction between the spatial and temporal patterns. To address this issue, inspired by the uniformity and alignment properties of contrastive learning techniques, we propose a novel framework called Session-based Recommendation with Spatio-Temporal Contrastive Learning Enhanced GNNs (RESTC). The idea is to supplement the GNN-based main supervised recommendation task with the temporal representation via an auxiliary cross-view contrastive learning mechanism. Furthermore, a novel global collaborative filtering graph (CFG) embedding is leveraged to enhance the spatial view in the main task. Extensive experiments demonstrate the significant performance of RESTC compared with the state-of-the-art baselines e.g., with an improvement as much as 27.08% gain on HR@20 and 20.10% gain on MRR@20. </p>

show abstract

Self-Supervised Hypergraph Convolutional Networks for Session-based Recommendation

Cited by 330 publications

References 35 publications

Session recommendation based on target attention and directed graph

Session recommendation based on target attention and directed graph

Efficient On-Device Session-Based Recommendation

Spatio-Temporal Contrastive Learning Enhanced GNNs for Session-based Recommendation

Contact Info

Product

Resources

About