Trust Relationship Prediction in Alibaba E-Commerce Platform

Gaofei Wang
Gaofei Wang
Yujie Qian
Yujie Qian
Chuizheng Meng
Chuizheng Meng
Zonghong Dai
Zonghong Dai

IEEE Transactions on Knowledge and Data Engineering, pp. 1024-1035, 2020.

Cited by: 1|Bibtex|Views306|Links
EI WOS
Keywords:
CorrelationFeature extractionBusinessTwitterGraphical modelsMore(1+)
Weibo:
Experimental results on four genres of real-world datasets show that the proposed method significantly outperforms comparison methods

Abstract:

This paper introduces how to infer trust relationships from billion-scale networked data to benefit Alibaba E-Commerce business. To effectively leverage the network correlations between labeled and unlabeled relationships to predict trust relationships, we formalize trust into multiple types and propose a graphical model to incorporate ty...More

Code:

Data:

0
Introduction
  • E-Commerce platform has led to a fundamental change in the way that businesses interact with their customers.
  • Almost all the famous platforms, such as Taobao1 and Amazon2, try to attract new customers or keep existing customers by developing sophisticated strategies to recommend products.
  • Traditional recommendations usually use contentbased, collaborative filtering-based or hybrid methods.
  • All these methods essentially categorize users/products into different groups and make recommendations based on the grouping information.
  • Leveraging the trust relationships between customers can significantly help E-Commerce.
  • This has been
Highlights
  • E-Commerce platform has led to a fundamental change in the way that businesses interact with their customers
  • 1. http://www.taobao.com, Alibaba E-Commerce platform 2. http://www.amazon.com 3. http://www.nielsen.com/us/en/insights/news/2013/under-theinfluence-consumer-trust-in-advertising.html demonstrated by social recommendation [11], [21], [31], [42], which suggests that the recommendation performance can be significantly improved with trust relationships
  • This paper studies how to infer trust relationships to facilitate Alibaba’s E-Commerce business
  • Experimental results on four genres of real-world datasets show that the proposed method significantly outperforms comparison methods
  • Our experimental results show that when dealing with large networks, eTrust-s can achieve > 2000× efficiency speedup, while guaranteeing a comparable accuracy to eTrust
  • A/B testings on Taobao product search/discovery further confirm the business value of the study. It shows that the performance of the proposed models is impacted when the network is too sparse, as the major improvement is caused by the network correlation features
Methods
  • EXISTING METHODS EXPLORATION

    the authors explore possible solutions and analyze their limitations.
  • Recent graph embedding methods such as DeepWalk (DW) [25] can be used to first learn an embedding vector for each user and calculate the trust score as dot product of two vectors
  • Supervised methods such as logistic regression (LR) [7] train a multi-label classifier to predict the label of a relationship based on the features extracted from heterogeneous user behaviors.
  • The labeled relationships in the datasets are divided into training and test set to learn and evaluate the comparison methods
Results
  • Evaluation Measures

    To quantitatively evaluate the proposed methods, the authors consider the following measurements.

    Accuracy Performance.
  • Recall, f1 and accuracy to evaluate the classification performance of the supervised/semi-supervised methods.
  • To evaluate the ranking performance of the unsupervised methods, the authors use precision at top k%, where k changes from 1 to 10 with interval 1 and from 10 to 100 with interval 10, as the evaluation measure, which is well-adopted to evaluate link prediction tasks [6], [18], [39], [40].
  • For Alibaba and Advogato with multiple labels, the authors evaluate the metric for each trust type separately
Conclusion
  • This paper studies how to infer trust relationships to facilitate Alibaba’s E-Commerce business.
  • The limitation of the proposed eTrust method is the potential inefficiency when dealing with large graphs.
  • The authors' experimental results show that when dealing with large networks, eTrust-s can achieve > 2000× efficiency speedup, while guaranteeing a comparable accuracy to eTrust.
  • It shows that the performance of the proposed models is impacted when the network is too sparse, as the major improvement is caused by the network correlation features.
  • The authors will study how to infer this multi-aspect trust in future works
Summary
  • Introduction:

    E-Commerce platform has led to a fundamental change in the way that businesses interact with their customers.
  • Almost all the famous platforms, such as Taobao1 and Amazon2, try to attract new customers or keep existing customers by developing sophisticated strategies to recommend products.
  • Traditional recommendations usually use contentbased, collaborative filtering-based or hybrid methods.
  • All these methods essentially categorize users/products into different groups and make recommendations based on the grouping information.
  • Leveraging the trust relationships between customers can significantly help E-Commerce.
  • This has been
  • Methods:

    EXISTING METHODS EXPLORATION

    the authors explore possible solutions and analyze their limitations.
  • Recent graph embedding methods such as DeepWalk (DW) [25] can be used to first learn an embedding vector for each user and calculate the trust score as dot product of two vectors
  • Supervised methods such as logistic regression (LR) [7] train a multi-label classifier to predict the label of a relationship based on the features extracted from heterogeneous user behaviors.
  • The labeled relationships in the datasets are divided into training and test set to learn and evaluate the comparison methods
  • Results:

    Evaluation Measures

    To quantitatively evaluate the proposed methods, the authors consider the following measurements.

    Accuracy Performance.
  • Recall, f1 and accuracy to evaluate the classification performance of the supervised/semi-supervised methods.
  • To evaluate the ranking performance of the unsupervised methods, the authors use precision at top k%, where k changes from 1 to 10 with interval 1 and from 10 to 100 with interval 10, as the evaluation measure, which is well-adopted to evaluate link prediction tasks [6], [18], [39], [40].
  • For Alibaba and Advogato with multiple labels, the authors evaluate the metric for each trust type separately
  • Conclusion:

    This paper studies how to infer trust relationships to facilitate Alibaba’s E-Commerce business.
  • The limitation of the proposed eTrust method is the potential inefficiency when dealing with large graphs.
  • The authors' experimental results show that when dealing with large networks, eTrust-s can achieve > 2000× efficiency speedup, while guaranteeing a comparable accuracy to eTrust.
  • It shows that the performance of the proposed models is impacted when the network is too sparse, as the major improvement is caused by the network correlation features.
  • The authors will study how to infer this multi-aspect trust in future works
Tables
  • Table1: Summarization of related methods. Notation tij is the trust score between user vi and vj, Vi denotes the neighbors of vi, f is a real-valued function to estimate the trust score of a relationship ei and yi is the label of ei
  • Table2: Statistics of dyadic and triadic correlation patterns
  • Table3: Notations
  • Table4: Dataset statistics. Notation |V |, |E| and |EL| denote the number of users, relationships and labeled relationships respectively
  • Table5: Efficiency performance(CPU time of model learning, s:seconds, m:minutes, h:hours)
  • Table6: Prediction performance of the supervised and semi-supervised methods on four datasets (%)
Download tables as Excel
Funding
  • This work is supported by Development Program of China (2016QY01W0200), National Key R&D Program of China (2018YFB1004401), NSFC for Distinguished Young Scholar (61825602) and NSFC under the grant No 61532021, REFERENCES [1] Black friday benchmark report: 2012 the year of the mobile shopper. https://www.ibm.com/developerworks/community/blogs/e874ec4d2a29-41a2-8fdd-16babe9d4d21/entry/Black Friday Benchmark Report 2012 the year of the mobile shopper?lang=zh, 2012. [2] S
Reference
  • Black friday benchmark report: 2012 the year of the mobile shopper. https://www.ibm.com/developerworks/community/blogs/e874ec4d2a29-41a2-8fdd-16babe9d4d21/entry/Black Friday Benchmark Report 2012 the year of the mobile shopper?lang=zh, 2012.
    Findings
  • S. Adalı. Modeling trust context in networks. Springer, 2013.
    Google ScholarFindings
  • L. A. Adamic and E. Adar. Friends and neighbors on the web. Social Networks, 25(3):211–230, 2003.
    Google ScholarLocate open access versionFindings
  • L. Backstrom and J. Leskovec. Supervised random walks: predicting and recommending links in social networks. In Proceedings of 4th ACM International Conference on Web Search and web Data Mining, pages 635–644, 2011.
    Google ScholarLocate open access versionFindings
  • D. Crandall, D. Cosley, D. Huttenlocher, J. Kleinberg, and S. Suri. Feedback effects between similarity and social influence in online communities. In Proceedings of the 14th ACM SIGKDD international conference on Knowledge discovery and data mining, pages 160–168, 2008.
    Google ScholarLocate open access versionFindings
  • Y. Dong, J. Zhang, J. Tang, N. V. Chawla, and B. Wang. Coupledlp: Link prediction in coupled networks. In Proceedings of the 21th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pages 199–208, 2015.
    Google ScholarLocate open access versionFindings
  • R.-E. Fan, K.-W. Chang, C.-J. Hsieh, X.-R. Wang, and C.-J. Lin. Liblinear: A library for large linear classification. Journal of Machine Learning Research, 9(Aug):1871–1874, 2008.
    Google ScholarLocate open access versionFindings
  • J. A. Golbeck. Computing and applying trust in web-based social networks. 2005.
    Google ScholarFindings
  • A. Goyal, F. Bonchi, and L. V. Lakshmanan. Learning influence probabilities in social networks. In Proceedings of 3rd ACM International Conference on Web Search and web Data Mining, pages 241–250, 2010.
    Google ScholarLocate open access versionFindings
  • R. Guha, R. Kumar, P. Raghavan, and A. Tomkins. Propagation of trust and distrust. In Proceedings of the 13th International Conference on World Wide Web, pages 403–412, 2004.
    Google ScholarLocate open access versionFindings
  • M. Jamali and M. Ester. A matrix factorization technique with trust propagation for recommendation in social networks. In Proceedings of the 4th ACM Conference on Recommender Systems, pages 135– 142, 2010.
    Google ScholarLocate open access versionFindings
  • T. N. Kipf and M. Welling. Semi-supervised classification with graph convolutional networks. In Proceedings of the 5th International Conference on Learning Representations, 2017.
    Google ScholarLocate open access versionFindings
  • N. Korovaiko and A. Thomo. Trust prediction from user-item ratings. Social Network Analysis and Mining, 3(3):749–759, 2013.
    Google ScholarLocate open access versionFindings
  • P. F. Lazarsfeld, R. K. Merton, et al. Friendship as a social process: A substantive and methodological analysis. Freedom and control in modern society, 18(1):18–66, 1954.
    Google ScholarLocate open access versionFindings
  • J. Leskovec, D. Huttenlocher, and J. Kleinberg. Predicting positive and negative links in online social networks. In Proceedings of the 19th International Conference on World Wide Web, pages 641–650, 2010.
    Google ScholarLocate open access versionFindings
  • J. Leskovec, D. Huttenlocher, and J. Kleinberg. Signed networks in social media. In Proceedings of the 28th SIGCHI Conference on Human Factors in Computing Systems, pages 1361–1370, 2010.
    Google ScholarLocate open access versionFindings
  • D. Liben-Nowell and J. M. Kleinberg. The link-prediction problem for social networks. Journal of American Society for Information Science and Technology, 58(7):1019–1031, 2007.
    Google ScholarLocate open access versionFindings
  • R. Lichtenwalter, J. T. Lussier, and N. V. Chawla. New perspectives and methods in link prediction. In Proceedings of the 16th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pages 243–252, 2010.
    Google ScholarLocate open access versionFindings
  • H. Liu, E.-P. Lim, H. W. Lauw, M.-T. Le, A. Sun, J. Srivastava, and Y. Kim. Predicting trusts among users of online communities: an epinions case study. In Proceedings of the 9th ACM Conference on Electronic Commerce, pages 310–319, 2008.
    Google ScholarLocate open access versionFindings
  • T. Lou, J. Tang, J. E. Hopcroft, Z. Fang, and X. Ding. Learning to predict reciprocity and triadic closure in social networks. ACM Transactions on Knowledge Discovery From Data, 7(2), 2013.
    Google ScholarLocate open access versionFindings
  • H. Ma, M. R. Lyu, and I. King. Learning to recommend with trust and distrust relationships. In Proceedings of the 3rd ACM Conference on Recommender Systems, pages 189–196, 2009.
    Google ScholarLocate open access versionFindings
  • P. MacCarron, K. Kaski, and R. Dunbar. Calling dunbar’s numbers. arXiv preprint arXiv:1604.02400, 2016.
    Findings
  • K. P. Murphy, Y. Weiss, and M. I. Jordan. Loopy belief propagation for approximate inference: An empirical study. In Proceedings of the 15th Conference on Uncertainty in Artificial Intelligence, pages 467–475, 1999.
    Google ScholarLocate open access versionFindings
  • V.-A. Nguyen, E.-P. Lim, J. Jiang, and A. Sun. To trust or not to trust? predicting online trusts using trust antecedent framework. In Proceedings of the 9th IEEE International Conference on Data Mining, pages 896–901, 2009.
    Google ScholarLocate open access versionFindings
  • B. Perozzi, R. Al-Rfou, and S. Skiena. Deepwalk: Online learning of social representations. In Proceedings of the 20th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pages 701–710, 2014.
    Google ScholarLocate open access versionFindings
  • R. D. Putnam. Bowling alone: The collapse and revival of American community. Simon and Schuster, 2001.
    Google ScholarLocate open access versionFindings
  • G. Salton and M. J. McGill. Introduction to Modern Information Retrieval.
    Google ScholarFindings
  • J. Tang, Y. Chang, and H. Liu. Mining social media with social theories: a survey. ACM SIGKDD Explorations Newsletter, 15(2):20–29, 2014.
    Google ScholarLocate open access versionFindings
  • J. Tang, H. Gao, X. Hu, and H. Liu. Exploiting homophily effect for trust prediction. In Proceedings of the 6th ACM International Conference on Web Search and Data Mining, pages 53–62, 2013.
    Google ScholarLocate open access versionFindings
  • J. Tang, H. Gao, H. Liu, and A. D. Sarma. etrust: Understanding trust evolution in an online world. In Proceedings of the 18th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, 2012.
    Google ScholarLocate open access versionFindings
  • J. Tang, X. Hu, and H. Liu. Social recommendation: a review. Social Network Analysis and Mining, 3(4):1113–1133, 2013.
    Google ScholarLocate open access versionFindings
  • J. Tang and H. Liu. Trust in social media. Synthesis Lectures on Information Security, Privacy, & Trust, 10(1):1–129, 2015.
    Google ScholarLocate open access versionFindings
  • J. Tang, T. Lou, J. Kleinberg, and S. Wu. Transfer learning to infer social ties across heterogeneous networks. ACM Transactions on Information Systems, 34(2):7, 2016.
    Google ScholarLocate open access versionFindings
  • J. Tang, J. Sun, C. Wang, and Z. Yang. Social influence analysis in large-scale networks. In Proceedings of the 15th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pages 807–816, 2009.
    Google ScholarLocate open access versionFindings
  • J. Tang, J. Zhang, L. Yao, J. Li, L. Zhang, and Z. Su. Arnetminer: extraction and mining of academic social networks. In SIGKDD’08, pages 990–998, 2008.
    Google ScholarLocate open access versionFindings
  • W. Tang, H. Zhuang, and J. Tang. Learning to infer social ties in large networks. In Joint European Conference on Machine Learning and Knowledge Discovery in Databases, pages 381–397, 2011.
    Google ScholarLocate open access versionFindings
  • P. Velickovic, G. Cucurull, A. Casanova, A. Romero, P. Lio, and Y. Bengio. Graph attention networks. In Proceedings of the 6th International Conference on Learning Representations, 2018.
    Google ScholarLocate open access versionFindings
  • C. Wang, V. Satuluri, and S. Parthasarathy. Local probabilistic models for link prediction. In Proceedings of the 7th IEEE International Conference on Data Mining, pages 322–331, 2007.
    Google ScholarLocate open access versionFindings
  • D. Wang, D. Pedreschi, C. Song, F. Giannotti, and A.-L. Barabasi. Human mobility, social ties, and link prediction. In Proceedings of the 17th ACM SIGKDD international conference on Knowledge discovery and pages=1100–1108, year=2011,.
    Google ScholarLocate open access versionFindings
  • Y. Yang, R. N. Lichtenwalter, and N. V. Chawla. Evaluating link prediction methods. Knowledge and Information Systems, 45(3):751–782, 2015.
    Google ScholarLocate open access versionFindings
  • D. Zhou, O. Bousquet, T. N. Lal, J. Weston, and B. Scholkopf. Learning with local and global consistency. In Advances in neural information processing systems, pages 321–328, 2004.
    Google ScholarLocate open access versionFindings
  • J. Zhu, H. Ma, C. Chen, and J. Bu. Social recommendation using low-rank semidefinite program. In Proceedings of the 25th Conference on Artificial Intelligence, pages 158–163, 2011.
    Google ScholarLocate open access versionFindings
  • X. Zhu, Z. Ghahramani, and J. D. Lafferty. Semi-supervised learning using gaussian fields and harmonic functions. In Proceedings of the 20th International conference on Machine learning, pages 912–919, 2003.
    Google ScholarLocate open access versionFindings
  • Gaofei Wang is working as the Staff Engineer in Alibaba Group. His interests span social tie prediction, information retrieval and vector retrieval. He received his Master degree from the Department of Computer Science and Technology, Harbin Engineering University in 2006. From then on, he has been working in Alibaba Group.
    Google ScholarLocate open access versionFindings
  • Yujie Qian is a PhD student at MIT Computer Science and Artificial Intelligence Laboratory, working with Prof. Regina Barzilay. He received his B.Eng. from the Department of Computer Science and Technology at Tsinghua University in 2017. His research interests include natural language processing and machine learning.
    Google ScholarFindings
  • Hongxia Yang is working as the Senior Staff Data Scientist and Director in Alibaba Group. She got her PhD degree in Statistics from Duke University in 2010. She has published over 30 over papers and held 9 filed/to be filed US patents and is serving as the associate editor for Applied Stochastic Models in Business and Industry.
    Google ScholarLocate open access versionFindings
Your rating :
0

 

Tags
Comments