基于图神经网络的权益推荐技术方案研究

doi:10.11959/j.issn.1000-0801.2023155

电信科学 ›› 2023, Vol. 39 ›› Issue (8): 91-101.doi: 10.11959/j.issn.1000-0801.2023155

• 研究与开发 • 上一篇

基于图神经网络的权益推荐技术方案研究

陶涛, 李珍, 王冀彬, 徐海勇, 江勇, 陈卓, 张润波, 胡清源

中移动信息技术有限公司，北京 100032

修回日期:2023-08-06 出版日期:2023-08-01 发布日期:2023-08-01
作者简介:陶涛（1972- ），男，博士，中移动信息技术有限公司副总经理、高级工程师，国务院政府特殊津贴获得者，主要研究方向为大数据领域的架构、系统规划、技术创新、运营管理等
李珍（1994- ），女，博士，中移动信息技术有限公司大数据平台部工程师，主要研究方向为多源数据融合、图计算等人工智能算法
王冀彬（1980- ），男，中移动信息技术有限公司大数据事业部总经理、高级工程师，主要研究方向为大数据与人工智能、数据资产及数据能力构建、数据服务和大数据生态建设等
徐海勇（1970- ），男，中移动信息技术有限公司总经理、正高级工程师，主要研究方向为移动通信、互联网、大数据
江勇（1973- ），男，中移动信息技术有限公司大数据事业部副总经理、高级工程师，主要研究方向为大数据与人工智能、多源数据融合、大数据产品及应用等
陈卓（1983- ），男，中移动信息技术有限公司大数据平台部副总经理、高级工程师，主要研究方向为大数据分布式计算、云原生大数据、海量数据处理和数据融合挖掘
张润波（1989- ），男，现就职于中移动信息技术有限公司，主要研究方向为机器学习、NLP、图数据库开发与应用、图计算、知识图谱等
胡清源（1989- ），男，现就职于中移动信息技术有限公司大数据平台部，主要研究方向为大数据中台能力、统筹全网数据管理、提升大数据治理技术等

Research on the graphical convolution neural network based benefits recommendation system strategy

Tao TAO, Zhen LI, Jibin WANG, Haiyong XU, Yong JIANG, Zhuo CEHN, Runbo ZHANG, Qingyuan HU

China Mobile Information Technology Center, Beijing 100032,China

Revised:2023-08-06 Online:2023-08-01 Published:2023-08-01

摘要/Abstract

摘要：

推荐系统是实现海量互联网权益产品智能化推荐的重要手段。为了提升个性化推荐的准确率，提出了基于图计算方法的深度学习推荐系统。针对用户行为数据存在多源异质的特性，基于深度学习图表示技术，对用户多维特征及权益产品之间的多种交互方式进行图结构化信息抽取及异质图建模，构建用户权益多元关系图谱，实现了各类交互信息（如用户—App、App—套餐、用户—套餐）的有效聚合。通过构建异质图卷积神经网络，学习各类异质性节点的高维特征向量，挖掘用户潜在偏好行为，提供具有较强可解释性的推荐链路，进而大幅提升推荐成功率并产生经济价值。

关键词: 异质图, 图卷积神经网络, 权益推荐, 多源数据

Abstract:

The recommendation system is one of the important methods to realize the intelligent recommendation of massive Internet benefit products.In order to improve the accuracy of personalized benefits recommendation, a deep learning recommendation system based on graph computing method was proposed.Considering the heterogeneity of multi-source data, a graph representation technology based on deep learning was carried out to construct the multiple relationship graph between users and benefit products.The multiple relationship graph extracted the information of graph structure, and model the heterogeneous graphs for the multi-dimensional features of users and the multiple interaction modes between rights and interests products, which effectively aggregated various interactive information and the multiple feature.A heterogeneous graph convolutional neural network was built to learn the high-dimensional feature vectors for various nodes, and excavate users＇ latent preferences to provide a recommendation link with strong interpretability, which greatly improved the recommendation success rate and generating economic value.

Key words: heterogeneous graph, graph convolutional neural network, benefit recommendation, multi-source data

中图分类号:

TP399

陶涛, 李珍, 王冀彬, 徐海勇, 江勇, 陈卓, 张润波, 胡清源. 基于图神经网络的权益推荐技术方案研究[J]. 电信科学, 2023, 39(8): 91-101.

Tao TAO, Zhen LI, Jibin WANG, Haiyong XU, Yong JIANG, Zhuo CEHN, Runbo ZHANG, Qingyuan HU. Research on the graphical convolution neural network based benefits recommendation system strategy[J]. Telecommunications Science, 2023, 39(8): 91-101.

图/表 9

图1

图2

图3

图4

图5

图6

图7

图8

图9

参考文献 16

[1]	AGRIMA M , POOJA G , SHARMA A K ,et al. A Study on recommender systems[J]. Smart and Sustainable Intelligent Systems, 2021.
[2]	SUN J N , ZHANG Y X , GUO W ,et al. Neighbor interaction aware graph convolution networks for recommendation[C]// Proceedings of the 43rd International ACM SIGIR Conference on Research and Development in Information Retrieval. New York:ACM Press, 2020: 1289-1298.
[3]	Curran Associates Inc. Probabilistic matrix factorization[R]. 2007.
[4]	CHENG Z Y , DING Y , HE X N ,et al. A^3NCF:an adaptive aspect attention model for rating prediction[C]// Proceedings of the Twenty-Seventh International Joint Conference on Artificial Intelligence. California:International Joint Conferences on Artificial Intelligence Organization, 2018.
[5]	柯福顺, 姚晓辉 . 基于多源数据融合的协同推荐方法[J]. 电信科学, 2015,31(7): 86-89.
	KE F S , YAO X H . Collaborating filtering method based on multiple data sources[J]. Telecommunications Science, 2015,31(7): 86-89.
[6]	YU X , REN X , SUN Y Z ,et al. Recommendation in heterogeneous information networks with implicit user feedback[C]// Proceedings of the 7th ACM conference on Recommender systems. New York:ACM Press, 2013: 347-350.
[7]	阳德青, 夏西, 叶琳 ,等. 知识驱动的推荐系统:现状与展望[J]. 信息安全学报, 2021,6(5): 35-51.
	YANG D Q , XIA X , YE L ,et al. Knowledge-enhanced recommender systems:a survey and prospect[J]. Journal of Cyber Security, 2021,6(5): 35-51.
[8]	KIPF T N , WELLING M . Semi-supervised classification with graph convolutional networks[J]. CoRR, 2016,abs/1609.02907.
[9]	HE X , ZHANG H , KAN M Y ,et al. Fast matrix factorization for online recommendation with implicit feedback[EB]. 2017:arXiv:1708.05024.
[10]	Curran Associates Inc. Reasoning with neural tensor networks for knowledge base completion[R]. 2013.
[11]	WU L , SUN P , FU Y ,et al. A neural influence diffusion model for social recommendation[EB]. 2019:arXiv:1904.10322.
[12]	WU L , LI J W , SUN P J ,et al. DiffNet++:a neural influence and interest diffusion network for social recommendation[J]. IEEE Transactions on Knowledge and Data Engineering, 2020 34(10): 4753-4766.
[13]	DEFFERRARD M , BRESSON X , VANDERGHEYNST P . Convolutional neural networks on graphs with fast localized spectral filtering[J]. CoRR, 2016,abs/1606.09375.
[14]	HAMILTON W L , YING R , LESKOVEC J ,et al. Inductive representation learning on large graphs[EB]. 2017:arXiv:1706.02216.
[15]	GORI M , MONFARDINI G , SCARSELLI F . A new model for learning in graph domains[C]// Proceedings of 2005 IEEE International Joint Conference on Neural Networks. Piscataway:IEEE Press, 2005: 729-734.
[16]	KUANR M , MOHAPATRA P . Assessment methods for evaluation of recommender systems:a survey[J]. Foundations of Computing and Decision Sciences, 2021,46(4): 393-421.

基于图神经网络的权益推荐技术方案研究

Research on the graphical convolution neural network based benefits recommendation system strategy

在线阅读

PDF下载

可视化

摘要/Abstract

引用本文

使用本文

图/表 9

参考文献 16

相关文章 1

Metrics

推荐阅读 0