《计算机应用》唯一官方网站 ›› 2024, Vol. 44 ›› Issue (6): 1760-1766.DOI: 10.11772/j.issn.1001-9081.2023060884
所属专题: 人工智能
收稿日期:
2023-07-07
修回日期:
2023-09-01
接受日期:
2023-09-05
发布日期:
2023-09-14
出版日期:
2024-06-10
通讯作者:
吕锡婷
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、互动创新基金资助:
Xiting LYU(), Jinghua ZHAO, Haiying RONG, Jiale ZHAO
Received:
2023-07-07
Revised:
2023-09-01
Accepted:
2023-09-05
Online:
2023-09-14
Published:
2024-06-10
Contact:
Xiting LYU
About author:
ZHAO Jinghua, born in 1984, Ph. D., associate professor. Her research interests include popularity prediction, interactive innovation.Supported by:
摘要:
针对在信息传播动态演化中,结构特征和时序特征以及两者间的交互表达难以有效捕获的问题,提出一种基于Transformer和关系图卷积网络的信息传播预测模型(TRGCN)。首先,构建由社交关系图和传播级联图组合而成的异构图,使用关系图卷积网络(RGCN)提取图中各节点的结构特征;其次,使用双向长短期记忆(Bi-LSTM)网络对各节点的时间嵌入重新编码,引入时间衰减项以不同的权重赋予不同时间位置的节点,获得节点的时序特征;最后,将结构特征和时序特征输入Transformer进行融合,得到时空特征以预测信息传播。在Twitter、Douban和Memetracker这3个真实数据集上的实验结果表明,相较于对比实验中的最优模型,TRGCN的Hits@100指标分别提升3.18%,5.96%和3.34%,Map@100指标分别提升11.60%,19.72%和8.47%,验证了所提模型的有效性和合理性。
中图分类号:
吕锡婷, 赵敬华, 荣海迎, 赵嘉乐. 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.
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.
数据集 | 用户数 | 用户社交 关系数 | 传播 级联数 | 传播级联的平均长度 |
---|---|---|---|---|
12 627 | 309 631 | 3 442 | 32.60 | |
Douban | 23 123 | 348 280 | 10 602 | 27.14 |
Memetracke | 4 709 | NULL | 12 661 | 16.24 |
表1 实验数据集统计数据
Tab. 1 Statistics of datasets used in experiments
数据集 | 用户数 | 用户社交 关系数 | 传播 级联数 | 传播级联的平均长度 |
---|---|---|---|---|
12 627 | 309 631 | 3 442 | 32.60 | |
Douban | 23 123 | 348 280 | 10 602 | 27.14 |
Memetracke | 4 709 | NULL | 12 661 | 16.24 |
参数 | 值 | 参数 | 值 |
---|---|---|---|
Batch Size | 16 | optimizer | Adam |
Learning Rate | 0.001 | n_layers | 2 |
Num Epoch | 50 | embed_dim | 64 |
Dropout Rate | 0.3 | n_heads | 8 |
表2 参数设置
Tab. 2 Parameters setting
参数 | 值 | 参数 | 值 |
---|---|---|---|
Batch Size | 16 | optimizer | Adam |
Learning Rate | 0.001 | n_layers | 2 |
Num Epoch | 50 | embed_dim | 64 |
Dropout Rate | 0.3 | n_heads | 8 |
数据集 | 模型 | Hits@10 | Hits@50 | Hits@100 | Map@10 | Map@50 | Map@100 |
---|---|---|---|---|---|---|---|
DeepDiffuse | 4.57 | 8.80 | 13.39 | 3.62 | 3.79 | 3.85 | |
Topo-LSTM | 6.51 | 15.48 | 23.68 | 4.31 | 4.67 | 4.79 | |
NDM | 21.52 | 32.23 | 38.31 | 14.30 | 14.80 | 14.89 | |
SNIDSA | 23.37 | 35.46 | 43.49 | 14.84 | 15.40 | 15.51 | |
FOREST | 26.18 | 40.95 | 50.39 | 17.21 | 17.88 | 18.02 | |
DyHGCN | 28.10 | 47.17 | 58.16 | 16.86 | 17.73 | 17.89 | |
TRGCN | 30.90 | 49.56 | 60.01 | 19.11 | 19.96 | 20.11 | |
Ddouban | DeepDiffuse | 9.02 | 14.93 | 19.13 | 4.80 | 5.07 | 5.13 |
Topo-LSTM | 9.16 | 14.94 | 18.93 | 5.00 | 5.26 | 5.32 | |
NDM | 10.31 | 18.87 | 24.02 | 5.54 | 5.93 | 6.00 | |
SNIDSA | 11.81 | 21.91 | 28.37 | 6.36 | 6.81 | 6.91 | |
FOREST | 14.16 | 24.79 | 31.25 | 7.89 | 8.38 | 8.47 | |
DyHGCN | 15.92 | 28.53 | 36.05 | 8.56 | 9.12 | 9.23 | |
TRGCN | 17.79 | 30.87 | 38.20 | 10.35 | 10.95 | 11.05 | |
Memetracker | DeepDiffuse | 13.93 | 26.50 | 34.77 | 8.14 | 8.69 | 8.80 |
NDM | 25.44 | 42.19 | 51.44 | 13.57 | 14.33 | 14.46 | |
FOREST | 29.43 | 47.41 | 56.77 | 16.37 | 17.21 | 17.34 | |
DyHGCN | 29.74 | 48.45 | 58.39 | 16.48 | 17.33 | 17.48 | |
TRGCN | 30.58 | 50.43 | 60.34 | 17.91 | 18.82 | 18.96 |
表3 不同数据集上模型的实验结果 (%)
Tab. 3 Experimental results of various models on different datasets
数据集 | 模型 | Hits@10 | Hits@50 | Hits@100 | Map@10 | Map@50 | Map@100 |
---|---|---|---|---|---|---|---|
DeepDiffuse | 4.57 | 8.80 | 13.39 | 3.62 | 3.79 | 3.85 | |
Topo-LSTM | 6.51 | 15.48 | 23.68 | 4.31 | 4.67 | 4.79 | |
NDM | 21.52 | 32.23 | 38.31 | 14.30 | 14.80 | 14.89 | |
SNIDSA | 23.37 | 35.46 | 43.49 | 14.84 | 15.40 | 15.51 | |
FOREST | 26.18 | 40.95 | 50.39 | 17.21 | 17.88 | 18.02 | |
DyHGCN | 28.10 | 47.17 | 58.16 | 16.86 | 17.73 | 17.89 | |
TRGCN | 30.90 | 49.56 | 60.01 | 19.11 | 19.96 | 20.11 | |
Ddouban | DeepDiffuse | 9.02 | 14.93 | 19.13 | 4.80 | 5.07 | 5.13 |
Topo-LSTM | 9.16 | 14.94 | 18.93 | 5.00 | 5.26 | 5.32 | |
NDM | 10.31 | 18.87 | 24.02 | 5.54 | 5.93 | 6.00 | |
SNIDSA | 11.81 | 21.91 | 28.37 | 6.36 | 6.81 | 6.91 | |
FOREST | 14.16 | 24.79 | 31.25 | 7.89 | 8.38 | 8.47 | |
DyHGCN | 15.92 | 28.53 | 36.05 | 8.56 | 9.12 | 9.23 | |
TRGCN | 17.79 | 30.87 | 38.20 | 10.35 | 10.95 | 11.05 | |
Memetracker | DeepDiffuse | 13.93 | 26.50 | 34.77 | 8.14 | 8.69 | 8.80 |
NDM | 25.44 | 42.19 | 51.44 | 13.57 | 14.33 | 14.46 | |
FOREST | 29.43 | 47.41 | 56.77 | 16.37 | 17.21 | 17.34 | |
DyHGCN | 29.74 | 48.45 | 58.39 | 16.48 | 17.33 | 17.48 | |
TRGCN | 30.58 | 50.43 | 60.34 | 17.91 | 18.82 | 18.96 |
模型 | Hits@10 | Hits@50 | Hits@100 | Map@10 | Map@50 | Map@100 |
---|---|---|---|---|---|---|
TRGCN | 30.90 | 49.56 | 60.01 | 19.11 | 19.96 | 20.11 |
-Social network | 30.65 | 49.10 | 59.70 | 18.31 | 19.13 | 19.29 |
-RGCN | 29.92 | 48.44 | 58.53 | 17.91 | 18.75 | 18.89 |
-Bi-LSTM | 29.82 | 48.67 | 59.55 | 18.13 | 18.99 | 19.15 |
-Time decay | 30.23 | 49.46 | 59.34 | 18.23 | 19.10 | 19.24 |
-Transformer | 25.02 | 42.08 | 52.97 | 14.66 | 15.42 | 15.57 |
表4 Twitter数据集上的消融实验结果 (%)
Tab. 4 Ablation experiment results on Twitter dataset
模型 | Hits@10 | Hits@50 | Hits@100 | Map@10 | Map@50 | Map@100 |
---|---|---|---|---|---|---|
TRGCN | 30.90 | 49.56 | 60.01 | 19.11 | 19.96 | 20.11 |
-Social network | 30.65 | 49.10 | 59.70 | 18.31 | 19.13 | 19.29 |
-RGCN | 29.92 | 48.44 | 58.53 | 17.91 | 18.75 | 18.89 |
-Bi-LSTM | 29.82 | 48.67 | 59.55 | 18.13 | 18.99 | 19.15 |
-Time decay | 30.23 | 49.46 | 59.34 | 18.23 | 19.10 | 19.24 |
-Transformer | 25.02 | 42.08 | 52.97 | 14.66 | 15.42 | 15.57 |
1 | YANG Y, TANG J, LEUNG C W, et al. RAIN: social role-aware information diffusion[C]// Proceedings of the 29th AAAI Conference on Artificial Intelligence. Palo Alto: AAAI Press, 2015: 367-373. |
2 | SUN L, RAO Y, ZHANG X, et al. MS-HGAT: memory enhanced sequential hypergraph attention network for information diffusion prediction [J]. Proceedings of the AAAI Conference on Artificial Intelligence, 2022, 36(4): 4156-4164. |
3 | GAO X, CAO Z, LI S, et al. Taxonomy and evaluation for microblog popularity prediction[J]. ACM Transactions on Knowledge Discovery from Data, 2019, 13(2): No.15. |
4 | 任敏捷,靳国庆,王晓雯,等. 基于XGBoost的微博流行度预测算法[J].数据采集与处理,2022,37(2):383-395. |
REN M J, JIN G Q, WANG X W, et al. Microblog popularity prediction algorithm based on XGBoost [J]. Journal of Data Acquisition and Processing, 2022, 37(2): 383-395. | |
5 | 陈稳,陈伟.基于计量指标多变量LSTM模型的新兴主题热度预测研究[J].数据分析与知识发现,2022,6(10):35-45. |
CHEN W, CHEN W. Predicting popularity of emerging topics with multivariable LSTM and bibliometric indicators [J]. Data Analysis and Knowledge Discovery, 2022, 6(10): 35-45. | |
6 | 张凤荔,王雪婷,王瑞锦,等. 融合动态图表示和自注意力机制的级联预测模型[J]. 电子科技大学学报, 2022, 51(1): 83-90. |
ZHANG F L, WANG X T, WANG R J, et al. Cascade prediction model based on dynamic graph representation and self-attention [J]. Journal of University of Electronic Science and Technology of China, 2022, 51(1): 83-90. | |
7 | SHEN H W, WANG D S, SONG C M, et al. Modeling and predicting popularity dynamics via reinforced poisson processes [J]. Proceedings of the AAAI Conference on Artificial Intelligence, 2014, 28(1): 291-297. |
8 | CAO Q, SHEN H, GAO H, et al. Predicting the popularity of online content with groupspecific models [C]// Proceedings of the 26th Conference on World Wide Web. Republic and Canton of Geneva, Switzerland: International World Wide Web Conferences Steering Committee, 2017: 765-766. |
9 | MISHRA S, M-A RIZOIU, XIE L. Feature driven and point process approaches for popularity prediction [C]// Proceedings of the 25th ACM International Conference on Information and Knowledge Management. New York: ACM, 2016: 1069-1078. |
10 | 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. | |
11 | 许阅, 刘光杰. 基于注意力机制的Bi-GRU内容流行度预测算法[J].电子测量技术, 2022, 45(3): 54-60. |
XU Y, LIU G J. Bi-GRU content popularity prediction algorithm based on attention mechanism[J]. Electronic Measurement Technology, 2022, 45(3): 54-60. | |
12 | YUAN C, LI J, ZHOU W, et al. DyHGCN: a dynamic heterogeneous graph convolutional network to learn users’ dynamic preferences for information diffusion prediction [C]// Proceedings of the 2020 Joint European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Database. Cham: Springer, 2020: 347-363. |
13 | 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. | |
14 | SCHLICHTKRULL M, KIPF T N, BLOEM P, et al. Modeling relational data with graph convolutional networks [C]// Proceedings of the 15th International Semantic Web Conference. Berlin: Springer, 2018: 593-607. |
15 | ISLAM M R, MUTHIAH S, ADHIKARI B,et al. DeepDiffuse: predicting the ‘who’ and ‘when’ in cascades [C]// Proceedings of the 2018 IEEE International Conference on Data Mining. Piscataway: IEEE, 2018: 1055-1060. |
16 | 孙加新,惠飞,张凯望, 等. 基于CNN-BiLSTM-AM模型的交通流量预测[J]. 计算机技术与发展, 2023, 33(2): 32-37,43. |
SUN J X, HUI F, ZHANG K W, et al. Traffic volume prediction based on CNN-BiLSTM-AM model [J]. Computer Technology and Development, 2023, 33(2): 32-37,43. | |
17 | 范伟,刘勇. 基于时空Transformer的社交网络信息传播预测[J]. 计算机研究与发展, 2022, 59(8): 1757-1769. |
FAN W, LIU Y. Social network information diffusion prediction based on spatial-temporal Transformer [J]. Journal of Computer Research and Development, 2022, 59(8): 1757-1769. | |
18 | WANG J, ZHANG V W, LIU Z,et al. Topological recurrent neural network for diffusion prediction[C]// Proceedings of the 2017 IEEE International Conference on Data Mining. Piscataway: IEEE, 2017: 475-484. |
19 | YANG C, SUN M, LIU H, et al. Neural diffusion model for microscopic cascade prediction [EB/OL]. [2023-07-02]. . |
20 | 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. |
21 | 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. Palo Alto: AAAI Press, 2019: 4033-4039. |
22 | LIU B, YANG D, SHI Y, et al. Improving information cascade modeling by social topology and dual role user dependency[C]// Proceedings of the 27th International Conference on Database Systems for Advanced Applications. Cham: Springer, 2022: 425-440. |
23 | CHEN X, ZHANG K, ZHOU F, et al. Information cascades modeling via deep multi-task learning [C]// Proceedings of the 42th International ACM SIGIR Conference on Research and Development in Information Retrieval. New York: ACM, 2018: 885-888. |
24 | 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. |
[1] | 任烈弘, 黄铝文, 田旭, 段飞. 基于DFT的频率敏感双分支Transformer多变量长时间序列预测方法[J]. 《计算机应用》唯一官方网站, 2024, 44(9): 2739-2746. |
[2] | 李金金, 桑国明, 张益嘉. APK-CNN和Transformer增强的多域虚假新闻检测模型[J]. 《计算机应用》唯一官方网站, 2024, 44(9): 2674-2682. |
[3] | 黄云川, 江永全, 黄骏涛, 杨燕. 基于元图同构网络的分子毒性预测[J]. 《计算机应用》唯一官方网站, 2024, 44(9): 2964-2969. |
[4] | 杨鑫, 陈雪妮, 吴春江, 周世杰. 结合变种残差模型和Transformer的城市公路短时交通流预测[J]. 《计算机应用》唯一官方网站, 2024, 44(9): 2947-2951. |
[5] | 方介泼, 陶重犇. 应对零日攻击的混合车联网入侵检测系统[J]. 《计算机应用》唯一官方网站, 2024, 44(9): 2763-2769. |
[6] | 贾洁茹, 杨建超, 张硕蕊, 闫涛, 陈斌. 基于自蒸馏视觉Transformer的无监督行人重识别[J]. 《计算机应用》唯一官方网站, 2024, 44(9): 2893-2902. |
[7] | 汪雨晴, 朱广丽, 段文杰, 李书羽, 周若彤. 基于交互注意力机制的心理咨询文本情感分类模型[J]. 《计算机应用》唯一官方网站, 2024, 44(8): 2393-2399. |
[8] | 丁宇伟, 石洪波, 李杰, 梁敏. 基于局部和全局特征解耦的图像去噪网络[J]. 《计算机应用》唯一官方网站, 2024, 44(8): 2571-2579. |
[9] | 邓凯丽, 魏伟波, 潘振宽. 改进掩码自编码器的工业缺陷检测方法[J]. 《计算机应用》唯一官方网站, 2024, 44(8): 2595-2603. |
[10] | 杨帆, 邹窈, 朱明志, 马振伟, 程大伟, 蒋昌俊. 基于图注意力Transformer神经网络的信用卡欺诈检测模型[J]. 《计算机应用》唯一官方网站, 2024, 44(8): 2634-2642. |
[11] | 石乾宏, 杨燕, 江永全, 欧阳小草, 范武波, 陈强, 姜涛, 李媛. 面向空气质量预测的多粒度突变拟合网络[J]. 《计算机应用》唯一官方网站, 2024, 44(8): 2643-2650. |
[12] | 李大海, 王忠华, 王振东. 结合空间域和频域信息的双分支低光照图像增强网络[J]. 《计算机应用》唯一官方网站, 2024, 44(7): 2175-2182. |
[13] | 刘瑞华, 郝子赫, 邹洋杨. 基于多层级精细特征融合的步态识别算法[J]. 《计算机应用》唯一官方网站, 2024, 44(7): 2250-2257. |
[14] | 黄梦源, 常侃, 凌铭阳, 韦新杰, 覃团发. 基于层间引导的低光照图像渐进增强算法[J]. 《计算机应用》唯一官方网站, 2024, 44(6): 1911-1919. |
[15] | 黎施彬, 龚俊, 汤圣君. 基于Graph Transformer的半监督异配图表示学习模型[J]. 《计算机应用》唯一官方网站, 2024, 44(6): 1816-1823. |
阅读次数 | ||||||
全文 |
|
|||||
摘要 |
|
|||||