ICCVAE: Item Concept Causal Variational Auto-Encoder for top-n recommendation DOI
Jingyun Feng, Qianqian Wang, Zhejun Huang

et al.

2022 7th International Conference on Intelligent Computing and Signal Processing (ICSP), Journal Year: 2023, Volume and Issue: unknown, P. 908 - 913

Published: April 21, 2023

Recently, recommendation systems, especially interpretable ones, have become increasingly popular. The system provides personalized recommendations to users based on their previous behaviour data. Existing approaches often transfer methods in disentangle learning, Variational Auto-Encoder (VAE) framework, into the system. However, VAE is proved sub-optimal due independence factor assumption. Unfortunately, few research focuses framework without an independent assumption for To escape from sub-optimality and provide more interpretability, this paper, we propose a new method top-n tasks called Item Concept Causal (ICCVAE), enabling build up causal structure factors, i.e., item concepts We conduct experiments prove superiority ICCVAE. Our can reach 5.3%, 3.4%, 6.1% definite improvement over baselines terms of Recall@20, Recall@50 NDCG@100.

Language: Английский

Graph neural networks DOI
Gabriele Corso, H. Stärk,

Stefanie Jegelka

et al.

Nature Reviews Methods Primers, Journal Year: 2024, Volume and Issue: 4(1)

Published: March 7, 2024

Language: Английский

Citations

52
Managing Metaverse Data Tsunami: Actionable Insights DOI Open Access
Bingxue Zhang, Gang Chen, Beng Chin Ooi

et al.

IEEE Transactions on Knowledge and Data Engineering, Journal Year: 2024, Volume and Issue: unknown, P. 1 - 20

Published: Jan. 1, 2024

In the metaverse physical space and virtual co-exist, interact simultaneously. While is virtually enhanced with information, continuously refreshed real-time, real-world information. To allow users to process manipulate information seamlessly between real digital spaces, novel technologies must be developed. These include smart interfaces, new augmented realities, efficient data storage, management, dissemination techniques. this paper, we first discuss some promising co-space applications. applications offer opportunities that neither of spaces can realize on its own. Then, further several emerging empower construction metaverse. After that, comprehensively centric challenges. Finally, envision what are likely required from database system perspectives.

Language: Английский

Citations

6
Graph Neural Network for Spatiotemporal Data: Methods and Applications DOI

Yun Li,

Dazhou Yu, Zhenke Liu

et al.

Published: Jan. 1, 2024

Abstract: In the era of big data, there has been a surge in availability data containing rich spatial and temporal information, offering valuable insights into dynamic systems processes for applications such as weather forecasting, natural disaster management, intelligent transport systems, precision agriculture. Graph neural networks (GNNs) have emerged powerful tool modeling understanding with dependencies to each other dependencies. There is large amount existing work that focuses on addressing complex spatiotemporal using GNNs. However, strong interdisciplinary nature created numerous GNNs variants specifically designed distinct application domains. Although techniques are generally applicable across various domains, cross-referencing these methods remains essential yet challenging due absence comprehensive literature review data. This article aims provide systematic overview technologies domain. First, ways constructing graphs from summarized help domain experts understand how generate types Then, categorization summary presented enable identify suitable support model developers advancing their research. Moreover, significant offered introduce broader range experts, assisting them exploring potential research topics enhancing impact work. Finally, open challenges future directions discussed.

Language: Английский

Citations

6
Self-supervised graph transformer networks for social recommendation DOI
Qinyao Li, Qimeng Yang, Shengwei Tian

et al.

Computers & Electrical Engineering, Journal Year: 2025, Volume and Issue: 123, P. 110121 - 110121

Published: Feb. 11, 2025

Language: Английский

Citations

0
Parallel graph neural architecture search optimization with incomplete features DOI
Haitao Yang, Zhaowei Liu, Dong Yang

et al.

Applied Soft Computing, Journal Year: 2025, Volume and Issue: unknown, P. 113068 - 113068

Published: April 1, 2025

Language: Английский

Citations

0
Heterogeneous information fusion based graph collaborative filtering recommendation DOI

Ruihui Mu,

Xiaoqin Zeng, Jiying Zhang

et al.

Intelligent Data Analysis, Journal Year: 2023, Volume and Issue: 27(6), P. 1595 - 1613

Published: Oct. 27, 2023

Nowadays, with the application of 5G, graph-based recommendation algorithms have become a research hotspot. Graph neural networks encode graph structure information in node representation through an iterative neighbor aggregation method, which can effectively alleviate problem data sparsity. In addition, more and be used collaborative filtering recommendation, such as user social graph, or item attributed etc. this paper, we propose novel heterogeneous fusion based models from different combines them together to enhance presentation learning. Through propagation aggregation, our model learn latent embeddings performance recommendation. Experimental results on datasets validate outperformance proposed framework.

Language: Английский

Citations

3
Deep latent representation enhancement method for social recommendation DOI
Xiaoyu Hou, Guobing Zou, Bofeng Zhang

et al.

Journal of Intelligent Information Systems, Journal Year: 2023, Volume and Issue: 62(1), P. 57 - 75

Published: July 29, 2023

Language: Английский

Citations

2
Diffusion social augmentation for social recommendation DOI

XiuBo Zang,

Hongbin Xia, Yuan Liu

et al.

The Journal of Supercomputing, Journal Year: 2024, Volume and Issue: 81(1)

Published: Nov. 25, 2024

Language: Английский

Citations

0
LASGRec: A Personalized Recommender Based on Learnable Attribute Sampling and Graph Neural Network DOI
Yufeng Wang, Xun Huang, Jianhua Ma

et al.

IEEE Transactions on Computational Social Systems, Journal Year: 2023, Volume and Issue: 11(2), P. 2930 - 2939

Published: Sept. 14, 2023

With the explosion of information, personalized recommender plays a vital role in almost all economic platforms. Usually, exploits user–item (UI) interactive data to learn users' latent interests, and then correspondingly conducts recommendations. To address problem sparse interactions, graph neural networks (GNNs) have been used efficiently representations users items, through structurally modeling inter-relationships among users, their attributes as graphs. However, most existing GNN-based methods ignore issue irrelevant attributes, which means that some an item are specific user's preference. Incorporating them into recommendation scheme may introduce noise decrease accuracy. Thus, above, this article proposes novel based on learnable attribute sampling heterogeneous network (LASGRec) improve recommender's performance. The work's contributions mainly threefold. First, history with user–item–attribute (UIA) is constructed, items sampled alleviate attributes. Second, using pruned UIA, GNN model appropriately items. Novelly, learnt embedding first aggregates user, these each incorporates two relationships: interacted its Finally, comprehensive experiments multiple real-world datasets demonstrate superiority proposed LASGRec over state-of-the-art deep (DNN-) schemes.

Language: Английский

Citations

0
ICCVAE: Item Concept Causal Variational Auto-Encoder for top-n recommendation DOI
Jingyun Feng, Qianqian Wang, Zhejun Huang

et al.

2022 7th International Conference on Intelligent Computing and Signal Processing (ICSP), Journal Year: 2023, Volume and Issue: unknown, P. 908 - 913

Published: April 21, 2023

Recently, recommendation systems, especially interpretable ones, have become increasingly popular. The system provides personalized recommendations to users based on their previous behaviour data. Existing approaches often transfer methods in disentangle learning, Variational Auto-Encoder (VAE) framework, into the system. However, VAE is proved sub-optimal due independence factor assumption. Unfortunately, few research focuses framework without an independent assumption for To escape from sub-optimality and provide more interpretability, this paper, we propose a new method top-n tasks called Item Concept Causal (ICCVAE), enabling build up causal structure factors, i.e., item concepts We conduct experiments prove superiority ICCVAE. Our can reach 5.3%, 3.4%, 6.1% definite improvement over baselines terms of Recall@20, Recall@50 NDCG@100.

Language: Английский

Citations

0