多任务推荐算法研究综述

doi:10.3778/j.issn.1673-9418.2303014

摘要/Abstract

摘要： 单任务推荐算法存在数据稀疏、冷启动和推荐效果不稳定等问题。多任务推荐算法可以将多种类型的用户行为数据和额外信息进行联合建模，从而更好地挖掘用户的兴趣和需求，以提高推荐效果和用户满意度，为解决单任务推荐算法存在的一系列问题提供了新思路。首先，梳理了多任务推荐算法的发展背景与趋势。其次，介绍了多任务推荐算法的实现步骤以及构建原则，并阐述了多任务学习具有数据增强、特征识别、特征互补和正则化效应等优势。然后，对不同共享模式的多任务学习方法在推荐算法中的应用进行了介绍，并对部分经典模型的优缺点及任务之间的关系进行了归纳总结。接着，介绍了多任务推荐算法常用的数据集和评估指标，并阐述了与其他推荐算法在数据集合评估指标方面的区别和联系。最后，指出多任务学习存在负迁移、参数优化冲突、可解释性差等不足，对多任务推荐算法与强化学习、凸函数优化方法、异构信息网络相结合进行了展望。

关键词: 推荐系统, 多任务学习, 多任务推荐

Abstract: Single-task recommendation algorithms have problems such as sparse data, cold start and unstable recommendation effect. Multi-task recommendation algorithms can jointly model multiple types of user behaviour data and additional information, to better explore the user’s interests and needs in order to improve the recommendation effect and user satisfaction, which provides a new way of thinking to solve a series of problems existing in single-task recommendation algorithms. Firstly, the development background and trend of multi-task recommendation algorithms are sorted out. Secondly, the implementation steps of the multi-task recommendation algorithm and the construction principle are introduced, and the advantages of multi-task learning with data enhancement, feature identification, feature complementation and regularization effect are elaborated. Then, the application of multi-task learning methods in recommendation algorithms with different sharing models is introduced, and the advantages and disadvantages of some classical models and the relationship between tasks are summarized. Then, the commonly used datasets and evaluation metrics for multi-task recommendation algorithms are introduced, and the differences and connections with other recommendation algorithms in terms of dataset evaluation metrics are elaborated. Finally, it is pointed out that multi-task learning has shortcomings such as negative migration, parameter optimization conflicts, poor interpretability, etc., and an outlook is given to the combination of multi-task recommendation algorithms with reinforcement learning, convex function optimization methods, and heterogeneous information networks.

Key words: recommendation system, multi-task learning, multi-task recommendation

中图分类号:

','1');return false;" target="_blank">

温民伟, 梅红岩, 袁凤源, 张晓宇, 张兴. 多任务推荐算法研究综述[J]. 计算机科学与探索, 2024, 18(2): 363-377.

WEN Minwei, MEI Hongyan, YUAN Fengyuan, ZHANG Xiaoyu, ZHANG Xing. Survey of Multi-task Recommendation Algorithms[J]. Journal of Frontiers of Computer Science and Technology, 2024, 18(2): 363-377.

参考文献

[1] DAS D, SAHOO L, DATTA S. A survey on recommendation system[J]. International Journal of Computer Applications, 2017, 160(7): 6-10.
[2] NUNES M, GERDING E, MCGROARTY F, et al. A comparison of multitask and single task learning with artificial neural networks for yield curve forecasting[J]. Expert Systems with Applications, 2019, 119: 362-375.
[3] NING X, KARYPIS G. Multi-task learning for recommender system[C]//Proceedings of the 2nd Asian Conference on Machine Learning, Japan, Nov 8-10, 2010: 269-284.
[4] VILALTA R, CARRIER C G, BRAZDIL P, et al. Inductive transfer[J]. Encyclopedia of Machine Learning, 2017, 1(1): 666-671.
[5] CARUANA R. Multitask learning[J]. Machine Learning, 1997, 28(1): 41-75.
[6] RUDER S. An overview of multi-task learning in deep neural networks[J]. arXiv:1706.05098, 2017.
[7] MA X, ZHAO L, HUANG G, et al. Entire space multi-task model: an effective approach for estimating post-click conversion rate[C]//Proceedings of the 41st International ACM SIGIR Conference on Research & Development in Information Retrieval. New York: ACM, 2018: 1137-1140.
[8] MA J, ZHAO Z, YI X, et al. Modeling task relationships in multi-task learning with multi-gate mixture-of-experts[C]//Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. New York: ACM, 2018: 1930-1939.
[9] TANG H, LIU J, ZHAO M, et al. Progressive layered extraction (PLE): a novel multi-task learning (MTL) model for personalized recommendations[C]//Proceedings of the 14th ACM Conference on Recommender Systems, New York, Sep 10, 2020. New York: ACM, 2020: 269-278.
[10] HUANG Z, RAO M, RAJU A, et al. MTL-SLT: multi-task learning for spoken language tasks[C]//Proceedings of the 4th Workshop on NLP for Conversational AI. Stroudsburg: ACL, 2022: 120-130.
[11] SAMPATH V, MAURTUA I, MARTíN J J A, et al. Attention guided multi-task learning for surface defect identification[J]. IEEE Transactions on Industrial Informatics, 2023, 19(9): 9713-9721.
[12] HU Y, CHEN C, LI R, et al. Gradient remedy for multi-task learning in end-to-end noise-robust speech recognition[J]. arXiv:2302.11362, 2023.
[13] SONG Y, WANG Y, WANG X, et al. Multi-task adaptive pooling enabled synergetic learning of RNA modification across tissue, type and species from low-resolution epitranscriptomes[J]. Briefings in Bioinformatics, 2023, 24.
[14] WEN H, ZHANG J, WANG Y, et al. Entire space multi-task modeling via post-click behavior decomposition for conversion rate prediction[C]//Proceedings of the 43rd International ACM SIGIR Conference on Research and Development in Information Retrieval. New York: ACM, 2020: 2377-2386.
[15] WANG Y, ZHANG J, DA Q, et al. Delayed feedback modeling for the entire space conversion rate prediction[J]. arXiv: 2011.11826, 2020.
[16] ZHANG W, BAO W, LIU X Y, et al. Large-scale causal approaches to debiasing post-click conversion rate estimation with multi-task learning[C]//Proceedings of the Web Conference 2020. New York: ACM, 2020: 2775-2781.
[17] WANG H, CHANG T W, LIU T, et al. ESCM2: entire space counterfactual multi-task model for post-click conversion rate estimation[J]. arXiv:2204.05125, 2022.
[18] ZHU F, ZHONG M, YANG X, et al. DCMT: a direct entire-space causal multi-task framework for post-click conversion estimation[J]. arXiv:2302.06141, 2023.
[19] JIN J, CHEN X, ZHANG W, et al. Multi-scale user behavior network for entire space multi-task learning[C]//Proceedings of the 31st ACM International Conference on Information and Knowledge Management, Atlanta, Oct 17, 2022. New York: ACM, 2022: 874-883.
[20] XI D, CHEN Z, YAN P, et al. Modeling the sequential dependence among audience multi-step conversions with multi-task learning in targeted display advertising[C]//Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery and Data Mining. New York: ACM, 2021: 3745-3755.
[21] ZHANG Y, LI X, YU Y, et al. Entire cost enhanced multi-task model for online-to-offline conversion rate prediction[C]//Proceedings of the 2022 Workshop on Deep Learning for Search and Recommendation, co-located with the 31st ACM International Conference on Information and Knowledge Management, Atlanta, Oct 17-21, 2022.
[22] BANSAL T, BELANGER D, MCCALLUM A. Ask the GRU: multi-task learning for deep text recommendations[C]//Proceedings of the 10th ACM Conference on Recommender Systems, Boston, Sep 7, 2016. New York: ACM, 2016: 107-114.
[23] YANG E, PAN J, WANG X, et al. AdaTask: a task-aware adaptive learning rate approach to multi-task learning[J]. arXiv:2211.15055, 2022.
[24] LIU J, LI X, AN B, et al. Multi-faceted hierarchical multi-task learning for recommender systems[C]//Proceedings of the 31st ACM International Conference on Information and Knowledge Management, Atlanta, Oct 17, 2022. New York: ACM, 2022: 3332-3341.
[25] LIN Z, YANG X, LIU S, et al. Personalized inter-task contrastive learning for CTR&CVR joint estimation[J]. arXiv: 2208.13442, 2022.
[26] MISRA I, SHRIVASTAVA A, GUPTA A, et al. Cross-stitch networks for multi-task learning[C]//Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition. Washington: IEEE Computer Society, 2016: 3994-4003.
[27] RUDER S, BINGEL J, AUGENSTEIN I, et al. Sluice networks: learning what to share between loosely related tasks[J]. arXiv:1705.08142, 2017.
[28] CHEN Z, WANG X, XIE X, et al. Co-attentive multi-task learning for explainable recommendation[C]//Proceedings of the 28th International Joint Conference on Artificial Intelligence, Macao, China, Aug 11-19, 2019: 2137-2143.
[29] XIN S, ESTER M, BU J, et al. Multi-task based sales predictions for online promotions[C]//Proceedings of the 28th ACM International Conference on Information and Knowledge Management. New York: ACM, 2019: 2823-2831.
[30] CHEN Y, YU J, ZHAO Y, et al. Task??s choice: pruning-based feature sharing (PBFS) for multi-task learning[J]. Entropy, 2022, 24(3): 432.
[31] SHAZEER N, MIRHOSEINI A, MAZIARZ K, et al. Outrageously large neural networks: the sparsely-gated mixture-of-experts layer[J]. arXiv:1701.06538, 2017.
[32] ZHAO Z, HONG L, WEI L, et al. Recommending what video to watch next: a multitask ranking system[C]//Proceedings of the 13th ACM Conference on Recommender Systems, Copenhagen, Sep 10, 2019. New York: ACM, 2019: 43-51.
[33] MA J, ZHAO Z, CHEN J, et al. SNR: sub-network routing for flexible parameter sharing in multi-task learning[C]//Proceedings of the 2019 AAAI Conference on Artificial Intelligence, Honolulu, Jan 27-Feb 1, 2019. Menlo Park: AAAI, 2019: 216-223.
[34] XU Z, ZHAO M, LIU L, et al. Mixture of virtual-kernel experts for multi-objective user profile modeling[C]//Proceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, Washington, Aug 14, 2022. New York: ACM, 2022: 4257-4267.
[35] WU X, MAGNANI A, CHAIDAROON S, et al. A multi-task learning framework for product ranking with BERT[C]//Proceedings of the ACM Web Conference 2022, Lyon, Apr 25, 2022. New York: ACM, 2022: 493-501.
[36] QIN Z, CHENG Y, ZHAO Z, et al. Multitask mixture of sequential experts for user activity streams[C]//Proceedings of the 26th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, Aug 20, 2020. New York: ACM, 2020: 3083-3091.
[37] WU H. MNCM: multi-level network cascades model for multi-task learning[C]//Proceedings of the 31st ACM International Conference on Information and Knowledge Management, Atlanta, Oct 17, 2022. New York: ACM, 2022: 4565-4569.
[38] XIN S, JIAO Y, LONG C, et al. Prototype feature extraction for multi-task learning[C]//Proceedings of the ACM Web Conference 2022, Lyon, Apr 25, 2022. New York: ACM, 2022: 2472-2481.
[39] TAO X, HA M, GUO X, et al. Task aware feature extraction framework for sequential dependence multi-task learning[J]. arXiv:2301.02494, 2023.
[40] LI D, ZHANG Z, YUAN S, et al. AdaTT: adaptive task-to-task fusion network for multitask learning in recommendations[J]. arXiv:2304.04959, 2023.
[41] SUN T, SHAO Y, LI X, et al. Learning sparse sharing architectures for multiple tasks[C]//Proceedings of the 2020 AAAI Conference on Artificial Intelligence. Menlo Park: AAAI, 2020: 8936-8943.
[42] XIAO X, CHEN H, LIU Y, et al. LT4REC: a lottery ticket hypothesis based multi-task practice for video recommendation system[J]. arXiv:2008.09872, 2020.
[43] DING K, DONG X, HE Y, et al. MSSM: a multiple-level sparse sharing model for efficient multi-task learning[C]//Proceedings of the 44th International ACM SIGIR Conference on Research and Development in Information Retrieval, Canada, Jul 11, 2021. New York: ACM, 2021: 2237-2241.
[44] CHEN X, GU X, FU L. Boosting share routing for multi-task learning[C]//Proceedings of the Companion of the Web Conference 2021, Ljubljana, Jun 3, 2021: 372-379.
[45] BAI T, XIAO Y, WU B, et al. A contrastive sharing model for multi-task recommendation[C]//Proceedings of the ACM Web Conference 2022. New York: ACM, 2022: 3239-3247.
[46] LIM N, HOOI B, NG S K, et al. Hierarchical multi-task graph recurrent network for next POI recommendation[C]//Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval, Madrid, Jul 7, 2022. New York: ACM, 2022: 1133-1143.
[47] WEN H, ZHANG J, LV F, et al. Hierarchically modeling micro and macro behaviors via multi-task learning for conversion rate prediction[C]//Proceedings of the 44th International ACM SIGIR Conference on Research and Development in Information Retrieval. New York: ACM, 2021: 2187-2191.
[48] VANSCHOREN J. Meta-learning[M]//Automated Machine Learning. Cham: Springer, 2019: 35-61.
[49] YANG K, LI X, LI J, et al. Meta-learning for recommendation system with the multi-tasking learning setting[C]//Proceedings of the 2019 IEEE 4th Advanced Information Technology, Electronic and Automation Control Conference, Chengdu, Dec 20-22, 2019. Piscataway: IEEE, 2020: 735-740.
[50] HE Y, FENG X, CHENG C, et al. MetaBalance: improving multi-task recommendations via adapting gradient magnitudes of auxiliary tasks[C]//Proceedings of the ACM Web Conference 2022. New York: ACM, 2022: 2205-2215.
[51] LEE S, SON Y. Multitask learning with single gradient step update for task balancing[J]. Neurocomputing, 2022, 467: 442-453.
[52] JAVALOY A, VALERA I. Rotograd: gradient homogenization in multitask learning[J]. arXiv:2103.02631, 2021.
[53] HE R, MCAULEY J. Ups and downs: modeling the visual evolution of fashion trends with one-class collaborative filtering[C]//Proceedings of the 25th International Conference on World Wide Web, Montreal, Apr 11-15, 2016: 507-517.
[54] HARPER F M, KONSTAN J A. The MovieLens datasets: history and context[J]. ACM Transactions on Interactive Intelligent Systems, 2015, 5(4): 1-19.
[55] 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.
[56] ORAMAS S, OSTUNI V C, NOIA T D, et al. Sound and music recommendation with knowledge graphs[J]. ACM Transactions on Intelligent Systems and Technology, 2016, 8(2): 1-21.
[57] SCHEDL M. The LFM-1b dataset for music retrieval and recommendation[C]//Proceedings of the 2016 ACM International Conference on Multimedia Retrieval, New York, Jun 6, 2016. New York: ACM, 2016: 103-110.
[58] KANG W C, MCAULEY J. Self-attentive sequential recommendation[C]//Proceedings of the 2018 IEEE International Conference on Data Mining. Piscataway: IEEE, 2018: 197-206.
[59] WANG X, HE X, WANG M, et al. Neural graph collaborative filtering[C]//Proceedings of the 42nd International ACM SIGIR Conference on Research and Development in Information Retrieval, Paris, Jul 18, 2019. New York: ACM, 2019: 165-174.
[60] FAWCETT T. An introduction to ROC analysis[J]. Pattern Recognition Letters, 2006, 27(8): 861-874.
[61] ZHU H, JIN J, TAN C, et al. Optimized cost per click in taobao display advertising[C]//Proceedings of the 23rd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. New York: ACM, 2017: 2191-2200.
[62] ZHANG Y, CHEN X. Explainable recommendation: a survey and new perspectives[J]. Foundations and Trends? in Information Retrieval, 2020, 14(1): 1-101.
[63] LU Y, FANG Y, SHI C. Meta-learning on heterogeneous information networks for cold-start recommendation[C]//Proceedings of the 26th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. New York: ACM, 2020: 1563-1573.
[64] LI H, WANG Y, LYU Z, et al. Multi-task learning for recommendation over heterogeneous information network[J]. IEEE Transactions on Knowledge and Data Engineering, 2022, 34(2): 789-802.
[65] LI P, WANG Z, REN Z, et al. Neural rating regression with abstractive tips generation for recommendation[C]//Proceedings of the 40th International ACM SIGIR Conference on Research and Development in Information Retrieval, Tokyo, Aug 7, 2017. New York: ACM, 2017: 345-354.
[66] VITHAYATHIL VARGHESE N, MAHMOUD Q H. A survey of multi-task deep reinforcement learning[J]. Electronics, 2020, 9(9): 1363.
[67] ZHANG Q, LIU J, DAI Y, et al. Multi-task fusion via reinforcement learning for long-term user satisfaction in recommender systems[C]//Proceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, Washington, Aug 14, 2022. New York: ACM, 2022: 4510-4520.
[68] ZHANG H, ZHAO P, XIAN X, et al. Click is not equal to purchase: multi-task reinforcement learning for multi-behavior recommendation[C]//Proceedings of the 2022 International Conference on Web Information Systems Engineering. Cham: Springer, 2022: 443-459.
[69] LIU X, LI L, HSIEH P C, et al. Developing multi-task recommendations with long-term rewards via policy distilled reinforcement learning[J]. arXiv:2001.09595, 2020.