Multiscale information diffusion prediction model based on hypergraph neural network

doi:10.11772/j.issn.1001-9081.2024111657

Journal of Computer Applications ›› 2025, Vol. 45 ›› Issue (11): 3529-3539.DOI: 10.11772/j.issn.1001-9081.2024111657

• Artificial intelligence • Previous Articles

Multiscale information diffusion prediction model based on hypergraph neural network

Jinghua ZHAO¹, Zhu ZHANG¹, Xiting LYU¹, Huidan LIN²()

^1.Business School，University of Shanghai for Science and Technology，Shanghai 200093，China
^2.Shanghai Innovation Center of Reverse Logistics and Supply Chain （Shanghai Polytechnic University），Shanghai 201209，China

Received:2024-11-22 Revised:2025-03-10 Accepted:2025-03-18 Online:2025-04-02 Published:2025-11-10
Contact: Huidan LIN
About author:ZHAO Jinghua， born in 1984， Ph. D.，associate professor. Her research interests include popularity prediction， interactive innovation.
ZHANG Zhu， born in 2002， M. S. candidate.His research interests include popularity prediction.
LYU Xiting， born in 1997， M. S. candidate. Her research interests include popularity prediction.
Supported by:
National Natural Science Foundation of China(72201173);Shanghai Educational Science Research Project(C2023292);Open Project of Shanghai Innovation Center of Reverse Logistics and Supply Chain

基于超图神经网络的多尺度信息传播预测模型

赵敬华¹, 张柱¹, 吕锡婷¹, 林慧丹²()

^1.上海理工大学管理学院，上海 200093
^2.上海市逆向物流与供应链协同创新中心（上海第二工业大学），上海 201209

通讯作者: 林慧丹
作者简介:赵敬华（1984—），女，山东冠县人，副教授，博士，主要研究方向：流行度预测、互动创新
张柱（2002—），男，湖南岳阳人，硕士研究生，主要研究方向：流行度预测
吕锡婷（1997—），女，浙江余姚人，硕士研究生，主要研究方向：流行度预测
基金资助:
国家自然科学基金资助项目(72201173);上海市教育科学研究项目(C2023292);上海市逆向物流与供应链协同创新中心开放课题

Abstract

Abstract:

To address the limitations of existing multiscale information diffusion prediction models， which ignore the dynamic characteristic of cascade propagation and exhibit limited performance in independent microscopic information prediction， a Multiscale Information Diffusion prediction model based on HyperGraph Neural Network （MIDHGNN）was proposed. Firstly， Graph Convolutional Network （GCN） was used to extract user social relationship features from the social network graphs， while HyperGraph Neural Network （HGNN）was used to extract global user preference features from propagation cascade graphs. These two types of features were fused to enable microscopic information diffusion prediction. Secondly， Gated Recurrent Unit （GRU） was employed to sequentially predict potential spreaders until reaching virtual users. The cumulative number of predicted users at each step was regarded as the determined cascade size for macroscopic propagation forecasting. Finally， a Reinforcement Learning （RL） framework using policy gradient to optimize parameters significantly enhanced macroscopic information diffusion prediction performance. For microscopic information diffusion prediction， compared to the suboptimal model， MIDHGNN achieves average improvements of 12.01%， 11.64%， and 9.74% in Hits@k on Twitter， Douban， and Android datasets， respectively， and average improvements of 31.31%， 14.85%， and 13.24% in mAP@k. For macroscopic prediction， MIDHGNN reduces the Mean Squared Logarithmic Error （MSLE） by at least 8.10%， 12.61%， and 3.24% on these three datasets， respectively， with all metrics significantly outperforming the comparison models， validating its effectiveness.

Key words: information diffusion prediction, Graph Convolutional Network (GCN), HyperGraph Neural Network (HGNN), Reinforcement Learning (RL), multiscale

摘要：

针对现有多尺度信息传播预测模型忽略了级联传播的动态性，以及独立进行微观信息预测时性能有待提高的问题，提出基于超图神经网络的多尺度信息传播预测模型（MIDHGNN）。首先，使用图卷积网络（GCN）提取社交网络图中蕴含的用户社交关系特征，使用超图神经网络（HGNN）提取传播级联图中蕴含的用户全局偏好特征，并融合这2类特征进行微观信息传播预测；其次，利用门控循环单元（GRU）连续预测传播用户，直至虚拟用户；再次，将每次预测所得用户总数作为级联的最终规模，完成宏观信息传播预测；最后，在模型中嵌入强化学习（RL）框架，采用策略梯度方法优化参数，提升宏观信息传播预测性能。在微观信息传播预测方面，相较于次优模型，MIDHGNN在Twitter、Douban、Android数据集上的Hits@k指标分别平均提升12.01%、11.64%、9.74%，mAP@k指标分别平均提升31.31%、14.85%、13.24%；在宏观预测方面，MIDHGNN在这3个数据集上的均方对数误差（MSLE）指标分别最少降低8.10%、12.61%、3.24%，各项指标均显著优于对比模型，验证了它的有效性。

关键词: 信息传播预测, 图卷积网络, 超图神经网络, 强化学习, 多尺度

CLC Number:

TP399

Jinghua ZHAO, Zhu ZHANG, Xiting LYU, Huidan LIN. Multiscale information diffusion prediction model based on hypergraph neural network[J]. Journal of Computer Applications, 2025, 45(11): 3529-3539.

赵敬华, 张柱, 吕锡婷, 林慧丹. 基于超图神经网络的多尺度信息传播预测模型[J]. 《计算机应用》唯一官方网站, 2025, 45(11): 3529-3539.

Figures/Tables 18

Fig. 1 Framework of MIDHGNN model

Fig. 2 Process of extracting users' global preference features by HGNN

Fig. 3 GRU model subgraph example

Fig. 4 Reinforcement learning model

Tab. 1 Statistical details of experimental datasets

数据集	#Users	#Links	#Cascades	Avg.Length
Twitter	12 627	309 631	3 442	32.60
Douban	23 123	348 280	10 602	27.14
Android	9 958	48 573	679	33.30

Tab. 2 Impact of time window length on model performance and training efficiency

$Δ$ /h	Hits@10	mAP@10	MSLE	训练时间/h
6	0.302	0.217	0.923	8.2
12	0.308	0.224	0.905	6.3
24	0.318	0.224	0.896	4.5
48	0.311	0.221	0.912	3.9

Tab. 2 Impact of time window length on model performance and training efficiency

$Δ$ /h	Hits@10	mAP@10	MSLE	训练时间/h
6	0.302	0.217	0.923	8.2
12	0.308	0.224	0.905	6.3
24	0.318	0.224	0.896	4.5
48	0.311	0.221	0.912	3.9

Tab. 3 Parameter settings

参数名	参数值	参数名	参数值
Batch Size	16	d_model	64
d_pos	8	warmup epochs	10
Num Epoch	50	embed_dim	64
Dropout Rate	0.1	k	5

Tab. 4 Experimental results of micro-scale prediction

数据集	模型	Hits@10	Hits@50	Hits@100	mAP@10	mAP@50	mAP@100
Twitter	NDM	0.215 2	0.322 3	0.383 1	0.143 0	0.148 0	0.148 9
	SNIDSA	0.233 7	0.354 6	0.434 9	0.148 4	0.154 0	0.155 1
	GraphSAGE	0.257 7	0.337 8	0.453 7	0.156 3	0.166 5	0.170 2
	FOREST	0.261 8	0.409 5	0.503 9	0.172 1	0.178 8	0.180 2
	TGAT	0.248 0	0.360 7	0.478 7	0.164 0	0.170 4	0.178 0
	MIDHGNN	0.318 4	0.443 7	0.534 5	0.227 9	0.233 1	0.236 3
Douban	NDM	0.103 1	0.188 7	0.240 2	0.055 4	0.059 3	0.060
	SNIDSA	0.118 1	0.219 1	0.283 7	0.063 6	0.068 1	0.069 1
	GraphSAGE	0.123 3	0.190 3	0.223 9	0.053 3	0.057 2	0.060 3
	FOREST	0.141 6	0.247 9	0.312 5	0.078 9	0.083 8	0.084 7
	TGAT	0.137 8	0.184 8	0.265 3	0.067 2	0.071 8	0.076 3
	MIDHGNN	0.160 2	0.275 8	0.345 4	0.091 1	0.096 0	0.097 0
Android	NDM	0.017 0	0.042 3	0.055 5	0.005 9	0.007 0	0.007 2
	GraphSAGE	0.065 5	0.094 3	0.118 2	0.052 2	0.054 1	0.057 3
	FOREST	0.086 6	0.173 9	0.231 4	0.062 8	0.066 7	0.067 5
	TGAT	0.071 2	0.166 2	0.196 4	0.061 1	0.063 2	0.066 1
	MIDHGNN	0.096 9	0.189 6	0.250 6	0.072 3	0.075 3	0.075 4

Tab. 5 Experimental results of macro-scale prediction

数据集	模型	MSLE
Twitter	DeepCas	2.261
	DeepHawkes	2.411
	FOREST	0.975
	TGAT	1.164
	MIDHGNN	0.896
Douban	DeepCas	2.122
	DeepHawkes	1.725
	FOREST	0.825
	TGAT	0.927
	MIDHGNN	0.721
Android	DeepCas	2.122
	DeepHawkes	1.971
	FOREST	0.556
	TGAT	0.741
	MIDHGNN	0.538

Tab. 6 Ablation experiment results on Twitter dataset

模型	Hits@100	mAP@100	MSLE
-GCN	0.515 4	0.220 2	0.935
-HGNN	0.503 2	0.214 0	1.021
-RL	0.520 2	0.223 0	—
-HGNN+RL	0.501 1	0.207 4	—
MIDHGNN	0.534 5	0.236 3	0.896

Fig. 5 Impact of user node embedding dimension on micro prediction index value

Tab. 7 Impact of user node embedding dimension on macro prediction index value

用户节点嵌入维度	MSLE	用户节点嵌入维度	MSLE
16	1.046	64	0.896
32	0.967	128	0.986

Fig. 6 Impact of position embedding dimension on micro prediction index value

Tab.8 Impact of position embedding dimension on macro prediction index value

位置嵌入维度	MSLE	位置嵌入维度	MSLE
2	1.134	16	1.037
4	1.083	32	1.042
8	0.896

Fig. 7 Impact of number of GRU layers on micro prediction index value

Tab.9 Impact of number of GRU layers on macro prediction index value

GRU层数	MSLE	GRU层数	MSLE
1	0.921	3	0.912
2	0.896	4	0.934

Fig. 8 Impact of number of attention heads on micro prediction index value

Tab.10 Impact of number of attention heads on macro prediction index value

注意力头数	MSLE	注意力头数	MSLE
2	0.945	10	0.901
4	0.091	12	0.899
6	0.903	14	0.904
8	0.896

References 39

[1]	王震宇，朱学芳.基于多模态Transformer的虚假新闻检测研究［J］.情报学报，2023，42（12）：1477-1486.
	WANG Z Y， ZHU X F. Research on fake news detection based on multimodal Transformer［J］. Journal of the China Society for Scientific and Technical Information， 2023， 42（12）： 1477-1486.
[2]	LI Q， XIE Y， WU X， et al. User behavior prediction model based on implicit links and multi-type rumor messages［J］. Knowledge-Based Systems， 2023， 262： No.110276.
[3]	WANG J， HU Y， JIANG T X， et al. Essential tensor learning for multimodal information-driven stock movement prediction［J］. Knowledge-Based Systems， 2023， 262： No.110262.
[4]	YANG X， YANG Y， SU J， et al. Who's next： rising star prediction via diffusion of user interest in social networks［J］. IEEE Transactions on Knowledge and Data Engineering， 2023， 35（5）： 5413-5425.
[5]	LI Y， JIN H， YU X， et al. Intelligent prediction of private information diffusion in social networks［J］. Electronics， 2020， 9（5）： No.719.
[6]	PORTILLO-VAN DIEST A， BALLESTER COMA L， MORTIER P， et al. Experience sampling methods for the personalised prediction of mental health problems in Spanish university students： protocol for a survey-based observational study within the PROMES-U project［J］. BMJ Open， 2023， 13（7）： No.e072641.
[7]	赵蓉英，李新来，李丹阳.高校新媒体信息传播网络结构及其演化特征［J］.情报科学，2022，40（6）：3-11.
	ZHAO R Y， LI X L， LI D Y. Structure and evolution feature on information transmission network of university new media［J］. Information Science， 2022， 40（6）： 3-11.
[8]	王晰巍，庄蕙荥，姜奕冰，等.重大突发事件下网络舆情传播中回声室网络结构研究［J］.情报理论与实践，2024，47（1）：101-109.
	WANG X W， ZHUANG H X， JIANG Y B， et al. Research on network structure of echo chamber in network public opinion propagation for major emergencies［J］. Information Studies： Theory and Application， 2024， 47（1）： 101-109.
[9]	CHEN X， ZHOU F， ZHANG K， et al. Information diffusion prediction via recurrent cascades convolution［C］// Proceedings of the IEEE 35th International Conference on Data Engineering. Piscataway： IEEE， 2019： 770-781.
[10]	SUN X， ZHOU J， LIU L， et al. Explicit time embedding based cascade attention network for information popularity prediction［J］. Information Processing and Management， 2023， 60（3）： No.103278.
[11]	苗琛香，刘小洋.融合超图注意力机制与图卷积网络的信息扩散预测［J］.计算机应用研究，2023，40（6）：1715-1720.
	MIAO C X， LIU X Y. Information diffusion prediction based on hypergraph attention mechanism and graph convolution network［J］. Application Research of Computers， 2023， 40（6）： 1715-1720.
[12]	GONG Y C， WANG M， LIANG W， et al. UHIR： an effective information dissemination model of online social hypernetworks based on user and information attributes［J］. Information Sciences， 2023， 644： No.119284.
[13]	丁晟春，包舟，刘笑迎.突发事件舆情传播中用户交互行为预测研究［J］.现代情报，2023，43（9）：111-123.
	DING S C， BAO Z， LIU X Y. Research on user behavior prediction of emergency public opinion communication［J］. Journal of Modern Information， 2023， 43（9）： 111-123.
[14]	JIA X， SHANG J， LIU D， et al. HeDAN： heterogeneous diffusion attention network for popularity prediction of online content［J］. Knowledge-Based Systems， 2022， 254： No.109659.
[15]	MENG F， CHEN L， HERRING P， et al. Information and disseminator features influences online negative information recognition and dissemination［J］. International Journal of Pattern Recognition and Artificial Intelligence， 2023， 37（3）： No.2350005.
[16]	赵小月，曾园园，江昊.基于C-DGCN的信息流行度预测［J］.武汉大学学报（工学版），2023，56（4）：506-514.
	ZHAO X Y， ZENG Y Y， JIANG H. Information popularity prediction model based on C-DGCN［J］. Engineering Journal of Wuhan University， 2023， 56（4）： 506-514.
[17]	ZHAO J H， ZHAO J L， FENG J. Information diffusion prediction based on cascade sequences and social topology［J］. Computers and Electrical Engineering， 2023， 109（Pt B）： No.108782.
[18]	梁少斌，陈志豪，魏晶晶，等.基于级联时空特征的信息传播预测方法［J］.模式识别与人工智能，2021，34（11）：969-978.
	LIANG S B， CHEN Z H， WEI J J， et al. Information diffusion prediction based on cascade spatial-temporal feature［J］. Pattern Recognition and Artificial Intelligence， 2021， 34（11）： 969-978.
[19]	吕锡婷，赵敬华，荣海迎，等.基于Transformer和关系图卷积网络的信息传播预测模型［J］.计算机应用，2024，44（6）：1760-1766.
	LYU X T， ZHAO J H， RONG H Y， et al. Information diffusion prediction model based on Transformer and relational graph convolutional network［J］. Journal of Computer Applications， 2024， 44（6）： 1760-1766.
[20]	WANG R， XU X， ZHANG Y. Multiscale information diffusion prediction with minimal substitution neural network［J］. IEEE Transactions on Neural Networks and Learning Systems， 2025， 36（1）： 1069-1080.
[21]	YANG C， WANG H， TANG J， et al. Full-scale information diffusion prediction with reinforced recurrent networks［J］. IEEE Transactions on Neural Networks and Learning Systems， 2023， 34（5）： 2271-2283.
[22]	KIPF T N， WELLING M. Semi-supervised classification with graph convolutional networks［EB/OL］. ［2024-09-26］..
[23]	操东林. 基于深度强化学习的金融投资研究［D］. 济南：山东财经大学， 2023.
	CAO D L. Research on financial investment based on deep reinforcement learning［D］. Jinan： Shandong University of Finance and Economics， 2023.
[24]	WILLIAMS R J. Simple statistical gradient-following algorithms for connectionist reinforcement learning［J］. Machine Learning， 1992， 8（3/4）： 229-256.
[25]	HODAS N O， LERMAN K. The simple rules of social contagion［J］. Scientific Reports， 2014， 4： No.4343.
[26]	ZHONG E， FAN W， WANG J， et al. ComSoc： adaptive transfer of user behaviors over composite social network［C］// Proceedings of the 18th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. New York： ACM， 2012： 696-704.
[27]	SANKAR A， ZHANG X， KRISHNAN A， et al. Inf-VAE： a variational autoencoder framework to integrate homophily and influence in diffusion prediction［C］// Proceedings of the 13th ACM International Conference on Web Search and Data Mining. New York： ACM， 2020： 510-518.
[28]	YANG C， SUN M， LIU H， et al. Neural diffusion model for microscopic cascade study［J］. IEEE Transactions on Knowledge and Data Engineering， 2018， 14（8）： 1128-1139.
[29]	WANG Z， CHEN C， LI W. A sequential neural information diffusion model with structure attention［C］// Proceedings of the 27th ACM International Conference on Information and Knowledge Management. New York： ACM， 2018： 1795-1798.
[30]	HAMILTON W L， YING R， LESKOVEC J. Inductive representation learning on large graphs［C］// Proceedings of the 31st International Conference on Neural Information Processing Systems. Red Hook： Curran Associates Inc.， 2017： 1025-1035.
[31]	LI C， MA J， GUO X， et al. DeepCas： an end-to-end predictor of information cascades［C］// Proceedings of the 26th International Conference on World Wide Web. Republic and Canton of Geneva： International World Wide Web Conferences Steering Committee， 2017： 577-586.
[32]	CAO Q， SHEN H， CEN K， et al. DeepHawkes： bridging the gap between prediction and understanding of information cascades［C］// Proceedings of the 2017 ACM Conference on Information and Knowledge Management. New York： ACM， 2017： 1149-1158.
[33]	YANG C， TANG J， SUN M， et al. Multi-scale information diffusion prediction with reinforced recurrent networks［C］// Proceedings of the 28th International Joint Conference on Artificial Intelligence. California： IJCAI.org， 2019： 4033-4039.
[34]	XU D， RUAN C， KORPEOGLU E， et al. Inductive representation learning on temporal graphs［EB/OL］. ［2024-07-06］. .
[35]	LIU B， YANG D， SHI Y， et al. Improving information cascade modeling by social topology and dual role user dependency［C］// Proceedings of the 2022 International Conference on Database Systems for Advanced Applications， LNCS 13245. Cham： Springer， 2022： 425-440.
[36]	CHEN X， ZHANG K， ZHOU F， et al. Information cascades modeling via deep multi-task learning［C］// Proceedings of the 42nd International ACM SIGIR Conference on Research and Development in Information Retrieval. New York： ACM， 2019： 885-888.
[37]	鲍鹏，徐昊.基于图注意力时空神经网络的在线内容流行度预测［J］.模式识别与人工智能，2019，32（11）：1014-1021.
	BAO P， XU H. Predicting popularity of online contents via graph attention based spatial-temporal neural network［J］. Pattern Recognition and Artificial Intelligence， 2019， 32（11）： 1014-1021.
[38]	JIAO P， CHEN H， BAO Q， et al. Enhancing multi-scale diffusion prediction via sequential hypergraphs and adversarial learning［C］// Proceedings of the 38th AAAI Conference on Artificial Intelligence. Palo Alto： AAAI Press， 2024： 8571-8581.
[39]	VASWANIA， SHAZEER N， PARMAR N， et al. Attention is all you need［C］// Proceedings of the 31st International Conference on Neural Information Processing Systems. Red Hook： Curran Associates Inc.， 2017： 6000-6010.

Multiscale information diffusion prediction model based on hypergraph neural network

基于超图神经网络的多尺度信息传播预测模型

RichHTML

PDF

Knowledge

Abstract

Cite this article

share this article

Figures/Tables 18

References 39

Related Articles 15

Recommended Articles

Metrics

[1]	Chao SHI, Yuxin ZHOU, Qian FU, Wanyu TANG, Ling HE, Yuanyuan LI. Action recognition algorithm for ADHD patients using skeleton and 3D heatmap [J]. Journal of Computer Applications, 2025, 45(9): 3036-3044.
[2]	Ziliang LI, Guangli ZHU, Yulei ZHANG, Jiajia LIU, Yixuan JIAO, Shunxiang ZHANG. Aspect-based sentiment analysis model integrating syntax and sentiment knowledge [J]. Journal of Computer Applications, 2025, 45(6): 1724-1731.
[3]	Quan WANG, Qixiang LU, Pei SHI. Multi-graph diffusion attention network for traffic flow prediction [J]. Journal of Computer Applications, 2025, 45(5): 1472-1479.
[4]	Yufei LONG, Yuchen MOU, Ye LIU. Multi-source data representation learning model based on tensorized graph convolutional network and contrastive learning [J]. Journal of Computer Applications, 2025, 45(5): 1372-1378.
[5]	Man CHEN, Xiaojun YANG, Huimin YANG. Pedestrian trajectory prediction based on graph convolutional network and endpoint induction [J]. Journal of Computer Applications, 2025, 45(5): 1480-1487.
[6]	Weichao DANG, Chujun SONG, Gaimei GAO, Chunxia LIU. Multi-behavior recommendation based on cascading residual graph convolutional network [J]. Journal of Computer Applications, 2025, 45(4): 1223-1231.
[7]	Kun FU, Shicong YING, Tingting ZHENG, Jiajie QU, Jingyuan CUI, Jianwei LI. Graph data augmentation method for few-shot node classification [J]. Journal of Computer Applications, 2025, 45(2): 392-402.
[8]	Pengcheng SONG, Lijun GUO, Rong ZHANG. Weakly supervised video anomaly detection with local-global temporal dependency [J]. Journal of Computer Applications, 2025, 45(1): 240-246.
[9]	Chuanlin PANG, Rui TANG, Ruizhi ZHANG, Chuan LIU, Jia LIU, Shibo YUE. Distributed power allocation algorithm based on graph convolutional network for D2D communication systems [J]. Journal of Computer Applications, 2024, 44(9): 2855-2862.
[10]	Guixiang XUE, Hui WANG, Weifeng ZHOU, Yu LIU, Yan LI. Port traffic flow prediction based on knowledge graph and spatio-temporal diffusion graph convolutional network [J]. Journal of Computer Applications, 2024, 44(9): 2952-2957.
[11]	Huanhuan LI, Tianqiang HUANG, Xuemei DING, Haifeng LUO, Liqing HUANG. Public traffic demand prediction based on multi-scale spatial-temporal graph convolutional network [J]. Journal of Computer Applications, 2024, 44(7): 2065-2072.
[12]	Shibin LI, Jun GONG, Shengjun TANG. Semi-supervised heterophilic graph representation learning model based on Graph Transformer [J]. Journal of Computer Applications, 2024, 44(6): 1816-1823.
[13]	Xiting LYU, Jinghua ZHAO, Haiying RONG, Jiale ZHAO. Information diffusion prediction model based on Transformer and relational graph convolutional network [J]. Journal of Computer Applications, 2024, 44(6): 1760-1766.
[14]	Longtao GAO, Nana LI. Aspect sentiment triplet extraction based on aspect-aware attention enhancement [J]. Journal of Computer Applications, 2024, 44(4): 1049-1057.
[15]	Xianfeng YANG, Yilei TANG, Ziqiang LI. Aspect-level sentiment analysis model based on alternating‑attention mechanism and graph convolutional network [J]. Journal of Computer Applications, 2024, 44(4): 1058-1064.