Sequential behavior recommendation based on user’s latent state and dependency learning

doi:10.11772/j.issn.1001-9081.2021101765

Abstract

Abstract:

At present， how to capture the dynamic changes and dependencies of user behaviors is an important problem in the field of sequential recommendation， which mainly faces challenges such as large behavior event space and complex sequential dependencies of behaviors. To address the above challenges， a sequential recommendation algorithm based on the learning of latent states of behavioral sequences and their dependency relationships was proposed. Firstly， the low-dimensional representation of the latent states of behavioral sequences was obtained by using the maximum pooling hierarchical structure. Then， the dependencies between the latent states were captured and described by graph neural network in order to achieve the learning of user behavior change patterns， which led to more accurate sequential recommendation effect. Experimental results show that compared with the recent Hierarchical Gating Network （HGN） baseline algorithm on the IPTV， New York City （NYC） and Tokyo （TKY） datasets， the proposed algorithm improves the performance evaluation metric recall by 30.03%， 29.48% and 33.75% respectively， and obtains 37.20%， 43.47% and 40.34% relative improvements on Normalized Discounted Cumulative Gain （NDCG） metric， respectively. And the ablation experimental results demonstrate the effectiveness of dependency learning of sequential states. Therefore， the proposed algorithm is especially suitable for solving the problems with sparse behaviors in single time slice and complex behavioral dependencies in sequential recommendation.

Key words: recommendation algorithm, sequential recommendation, sequential behavior, behavior prediction, latent state learning

摘要：

如何捕捉用户行为的动态变化及依赖关系是当前时序推荐领域的一个重要问题，主要面临着行为事件空间庞大、行为的时序依赖关系复杂等挑战。针对以上挑战，提出了一种基于行为序列潜在状态及其依赖关系学习的时序推荐算法。首先，利用最大池化层级结构获得行为序列潜在状态的低维表征；然后，通过图神经网络捕捉和描述潜在状态之间的依赖关系以实现用户行为变化模式的学习，从而获得更准确的时序推荐效果。实验结果表明，所提算法在节目点播（IPTV）、纽约（NYC）和东京（TKY）这3个数据集上与近年的分层门控网络（HGN）基线算法相比，在性能评估指标召回率上分别提高了30.03%、29.48%和33.75%，在归一化折损累计增益（NDCG）指标上分别获得了37.20%、43.47%和40.34%的相对提升，且消融实验结果表明了时序状态的依赖关系学习的有效性，因此所提算法尤其适用于解决时序推荐中单一时间片中行为稀疏以及行为依赖关系复杂的问题。

关键词: 推荐算法, 时序推荐, 时序行为, 行为预测, 潜在状态学习

CLC Number:

TP183

Wen WEN, Fangyu LIANG. Sequential behavior recommendation based on user’s latent state and dependency learning[J]. Journal of Computer Applications, 2022, 42(12): 3756-3762.

温雯, 梁方宇. 基于用户潜在状态及依赖关系学习的时序行为推荐[J]. 《计算机应用》唯一官方网站, 2022, 42(12): 3756-3762.

Figures/Tables 5

Tab. 1 Description of symbols

符号	说明
$U, ε$	用户集合，物品（事件）集合
$e$	表示物品e的序列号
$X t u$ 或 $X t$	表示第t时期的用户-物品交互矩阵
$x t j$	第t个时期中第j个时段的用户-物品交互向量
$E$	用户（物品）参数矩阵
$v u$	用户特征向量
$v e$	物品（事件）特征向量
$z u, t j$	用户在第t个时期中第j个时段的潜在状态
$z ˜ u, t j$	用户在第t个时期中第j个时段的预测潜在状态
$Φ$	算法学习参数
$τ$	一个时期的长度
d	潜在状态的维度大小

Tab. 1 Description of symbols

符号	说明
$U, ε$	用户集合，物品（事件）集合
$e$	表示物品e的序列号
$X t u$ 或 $X t$	表示第t时期的用户-物品交互矩阵
$x t j$	第t个时期中第j个时段的用户-物品交互向量
$E$	用户（物品）参数矩阵
$v u$	用户特征向量
$v e$	物品（事件）特征向量
$z u, t j$	用户在第t个时期中第j个时段的潜在状态
$z ˜ u, t j$	用户在第t个时期中第j个时段的预测潜在状态
$Φ$	算法学习参数
$τ$	一个时期的长度
d	潜在状态的维度大小

Tab. 2 Statistics of datasets used in experiments

数据集	用户数	物品数	历史数	划分时间戳
IPTV	2 920	32 341	2 227 811	Nov 2， 2015
NYC	1 083	38 333	227 428	Jan 1， 2013
TKY	2 293	61 858	573 703	Jan1， 2013

Tab. 3 Performance comparison with baseline algorithms

基线算法	IPTV			NYC			TKY
基线算法	Recall@5	NDCG@5	MAP@5	Recall@5	NDCG@5	MAP@5	Recall@5	NDCG@5	MAP@5
提高比率/%	30.034 4	37.196 8	40.974 4	5.038 5	4.529 7	3.915 2	7.997 6	6.247 9	4.724 4
Mult-VAE	0.112 2	0.177 9	0.168 8	0.133 9	0.119 9	0.101 4	0.125 8	0.119 1	0.096 7
HRM	0.239 0	0.169 8	0.142 6	0.320 7	0.266 9	0.220 6	0.297 7	0.246 4	0.194 1
Caser	0.217 7	0.145 6	0.118 0	0.325 9	0.271 6	0.224 8	0.314 6	0.266 4	0.213 7
HGN	0.291 0	0.222 6	0.195 0	0.273 7	0.210 7	0.166 1	0.272 6	0.227 8	0.179 4
TIFU-KNN	0.253 3	0.203 5	0.180 2	0.337 4	0.289 2	0.245 2	0.337 6	0.300 9	0.254 0
本文算法	0.3784	0.3054	0.2749	0.3544	0.3023	0.2548	0.3646	0.3197	0.2660

Fig. 1 Results of ablation experiments

Fig. 2 Sensitivity analysis of hyperparameters

References 27

1	HAN J Y， ZHENG L， XU Y B， et al. Adaptive deep modeling of users and items using side information for recommendation［J］. IEEE Transactions on Neural Networks and Learning Systems， 2020， 31（3）：737-748. 10.1109/tnnls.2019.2909432
2	刘华真，王巍，谷壬倩，等. 基于用户浏览行为的个性化推荐研究综述［J］. 计算机应用研究， 2021， 38（8）：2268-2277. 10.19734/j.issn.1001-3695.2020.10.0347
	LIU H Z， WANG W， GU R Q， et al. Survey of personalized recommendation study based on user browsing behavior［J］. Application Research of Computers， 2021， 38（8）：2268-2277. 10.19734/j.issn.1001-3695.2020.10.0347
3	YANG D Q， ZHANG D Q， ZHENG V W， et al. Modeling user activity preference by leveraging user spatial temporal characteristics in LBSNs［J］. IEEE Transactions on Systems， Man， and Cybernetics： Systems， 2015， 45（1）：129-142. 10.1109/tsmc.2014.2327053
4	WANG W Q， YIN H Z， DU X Z， et al. TPM： a temporal personalized model for spatial item recommendation［J］. ACM Transactions on Intelligent Systems and Technology， 2018， 9（6）： No.61. 10.1145/3230706
5	SALAKHUTDINOV R， MNIH A. Probabilistic matrix factorization［C］// Proceedings of the 20th International Conference on Neural Information Processing Systems. Red Hook， NY： Curran Associates Inc.， 2007： 1257-1264. 10.1145/1390156.1390267
6	KOREN Y， BELL R， VOLINSKY C. Matrix factorization techniques for recommender systems［J］. Computer， 2009， 42（8）：30-37. 10.1109/mc.2009.263
7	BARKAN O， KOENIGSTEIN N. Item2Vec： neural item embedding for collaborative filtering ［C］// Proceedings of the IEEE 26th International Workshop on Machine Learning for Signal Processing. Piscataway： IEEE， 2016：1-6. 10.1109/mlsp.2016.7738886
8	LIANG D W， KRISHNAN R G， HOFFMAN M D， et al. Variational autoencoders for collaborative filtering［C］// Proceedings of the 2018 World Wide Web Conference. Republic and Canton of Geneva： International World Wide Web Conferences Steering Committee， 2018： 689-698. 10.1145/3178876.3186150
9	KOREN Y. Collaborative filtering with temporal dynamics［C］// Proceedings of the 15th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. New York： ACM， 2009： 447-456. 10.1145/1557019.1557072
10	RENDLE S， FREUDENTHALER C， SCHMIDT-THIEME L. Factorizing personalized Markov chains for next-basket recommendation［C］// Proceedings of the 19th International Conference on World Wide Web. New York： ACM， 2010： 811-820. 10.1145/1772690.1772773
11	HE R N， McAULEY J. Fusing similarity models with Markov chains for sparse sequential recommendation［C］// Proceedings of the IEEE 16th International Conference on Data Mining. Piscataway： IEEE， 2016： 191-200. 10.1109/icdm.2016.0030
12	HIDASI B， KARATZOGLOU A， BALTRUNAS L， et al. Session-based recommendations with recurrent neural networks［EB/OL］. （2016-03-29）［2021-10-14］..
13	RANGAPURAM S S， SEEGER M， GASTHAUS J， et al. Deep state space models for time series forecasting［C］// Proceedings of the 32nd International Conference on Neural Information Processing Systems. Red Hook， NY： Curran Associates Inc.， 2018： 7796-7805.
14	LI L Y， YAN J C， YANG X K， et al. Learning interpretable deep state space model for probabilistic time series forecasting［C］// Proceedings of the 28th International Joint Conference on Artificial Intelligence. California： ijcai.org， 2019： 2901-2908. 10.24963/ijcai.2019/402
15	KIPF T， van der POL E， WELLING M. Contrastive learning of structured world models［EB/OL］. （2020-01-05）［2021-10-14］..
16	TANG J X， WANG K. Personalized top-N sequential recommendation via convolutional sequence embedding［C］// Proceedings of the 11th ACM International Conference on Web Search and Data Mining. New York： ACM， 2018： 565-573. 10.1145/3159652.3159656
17	MA C， KANG P， LIU X. Hierarchical gating networks for sequential recommendation［C］// Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. New York： ACM， 2019： 825-833. 10.1145/3292500.3330984
18	HU H J， HE X N， GAO J Y， et al. Modeling personalized item frequency information for next-basket recommendation［C］// Proceedings of the 43rd International ACM SIGIR Conference on Research and Development in Information Retrieval. New York： ACM， 2020： 1071-1080. 10.1145/3397271.3401066
19	王娜，何晓明，刘志强，等. 一种基于用户播放行为序列的个性化视频推荐策略［J］. 计算机学报， 2020， 43（1）：123-135. 10.11897/SP.J.1016.2020.00123
	WANG N， HE X M， LIU Z Q， et al. Personalized video recommendation strategy based on user’s playback behavior sequence［J］. Chinese Journal of Computers， 2020， 43（1）：123-135. 10.11897/SP.J.1016.2020.00123
20	孙光福，吴乐，刘淇，等. 基于时序行为的协同过滤推荐算法［J］. 软件学报， 2013， 24（11）：2721-2733. 10.3724/sp.j.1001.2013.04478
	SUN G F， WU L， LIU Q， et al. Recommendations based on collaborative filtering by exploiting sequential behaviors［J］. Journal of Software， 2013， 24（11）：2721-2733. 10.3724/sp.j.1001.2013.04478
21	YIN D W， HONG L J， XUE Z Z， et al. Temporal dynamics of user interests in tagging systems［C］// Proceedings of the 25th AAAI Conference on Artificial Intelligence. Palo Alto， CA： AAAI Press， 2011：1279-1285. 10.1609/aaai.v25i1.8095
22	ZHOU C， BAI J Z， SONG J S， et al. ATRank： an attention-based user behavior modeling framework for recommendation［C］// Proceedings of the 32nd AAAI Conference on Artificial Intelligence. Palo Alto， CA： AAAI Press， 2018：4564-4571. 10.1609/aaai.v32i1.11618
23	蔡海尼，牛冰慧，文俊浩，等. 基于时序模型和矩阵分解的推荐算法［J］. 计算机应用研究， 2018， 35（6）：1624-1627， 1659. 10.3969/j.issn.1001-3695.2018.06.005
	CAI H N， NIU B H， WEN J H， et al. Recommender algorithm based on time series model and matrix factorization［J］. Application Research of Computers， 2018， 35（6）：1624-1627， 1659. 10.3969/j.issn.1001-3695.2018.06.005
24	YE W W， WANG S Q， CHEN X， et al. Time matters： sequential recommendation with complex temporal information［C］// Proceedings of the 43rd International ACM SIGIR Conference on Research and Development in Information Retrieval. New York： ACM， 2020： 1459-1468. 10.1145/3397271.3401154
25	WANG P F， GUO J F， LAN Y Y， et al. Learning hierarchical representation model for next basket recommendation［C］// Proceedings of the 38th International ACM SIGIR conference on Research and Development in Information Retrieval. New York： ACM， 2015：403-412. 10.1145/2766462.2767694
26	KINGMA D P， BA J L. Adam： a method for stochastic optimization［EB/OL］. （2017-01-30）［2021-10-14］..
27	HINTON G E， SRIVASTAVA N， KRIZHEVSKY A， et al. Improving neural networks by preventing co-adaptation of feature detectors［EB/OL］. （2012-07-03）［2021-10-14］..

[1]	Xingyao YANG, Hongtao SHEN, Zulian ZHANG, Jiong YU, Jiaying CHEN, Dongxiao WANG. Sequential recommendation based on hierarchical filter and temporal convolution enhanced self-attention network [J]. Journal of Computer Applications, 2024, 44(10): 3090-3096.
[2]	Runchao PAN, Qishan YU, Hongfei XIONG, Zhihui LIU. Collaborative recommendation algorithm based on deep graph neural network [J]. Journal of Computer Applications, 2023, 43(9): 2741-2746.
[3]	Yu LING, Zhilong SHAN. Knowledge concept recommendation system based on interest enhancement [J]. Journal of Computer Applications, 2023, 43(12): 3697-3702.
[4]	Lei TIAN, Lina GE. Advertising recommendation algorithm based on differential privacy [J]. Journal of Computer Applications, 2023, 43(11): 3346-3350.
[5]	Xiaohan YANG, Guosheng HAO, Xiehua ZHANG, Zihao YANG. Collaborative filtering algorithm based on collaborative training and Boosting [J]. Journal of Computer Applications, 2023, 43(10): 3136-3141.
[6]	Tengyue HAN, Shaozhang NIU, Wen ZHANG. Multimodal sequential recommendation algorithm based on contrastive learning [J]. Journal of Computer Applications, 2022, 42(6): 1683-1688.
[7]	Meng YU, Wentao HE, Xuchuan ZHOU, Mengtian CUI, Keqi WU, Wenjie ZHOU. Review of recommendation system [J]. Journal of Computer Applications, 2022, 42(6): 1898-1913.
[8]	Suqi ZHANG, Xinxin WANG, Shiyao SHE, Junhua GU. Knowledge graph recommendation model with multiple time scales and feature enhancement [J]. Journal of Computer Applications, 2022, 42(4): 1093-1098.
[9]	YANG Li, WANG Shihui, ZHU Bo. Point-of-interest recommendation algorithm combing dynamic and static preferences [J]. Journal of Computer Applications, 2021, 41(2): 398-406.
[10]	ZHU Simiao, Wei Shiwei, WEI Siheng, YU Dunhui. Video recommendation algorithm based on danmaku sentiment analysis and topic model [J]. Journal of Computer Applications, 2021, 41(10): 2813-2819.
[11]	TIAN Baojun, LIU Shuang, FANG Jiandong. Hybrid recommendation algorithm by fusion of topic information and convolution neural network [J]. Journal of Computer Applications, 2020, 40(7): 1901-1907.
[12]	ZHANG Junru, ZHAO Xiaoyan, YUAN Peiyan. Federated security tree algorithm for user privacy protection [J]. Journal of Computer Applications, 2020, 40(10): 2980-2985.
[13]	LI Bo, ZHANG Xiao, YAN Jingyi, LI Kewei, LI Heng, LING Yulong, ZHANG Yong. Application of KNN algorithm based on value difference metric and clustering optimization in bank customer behavior prediction [J]. Journal of Computer Applications, 2019, 39(9): 2784-2788.
[14]	WANG Lei, REN Hang, GONG Kai. Social recommendation method based on multi-dimensional trust and collective matrix factorization [J]. Journal of Computer Applications, 2019, 39(5): 1269-1274.
[15]	FAN Wei, XIE Cong, XIAO Chunjing, CAO Shuyan. Random walking recommendation algorithm based on combinational category space [J]. Journal of Computer Applications, 2019, 39(4): 984-988.