Collaborative QoS prediction with context-sensitive matrix factorization

https://doi.org/10.1016/j.future.2017.06.020Get rights and content

Highlights

  • This paper presents a general context-sensitive approach for collaborative QoS prediction.

  • Interactions of users-to-services and environment-to-environment are considered simultaneously as contextual factors in the QoS data.

  • Our method takes advantages of both implicit and explicit factors entailed in the QoS data through exploiting contextual information.

  • Experimental results reflect that this study offers an efficient global optimization, enabling robust and accurate prediction results.

Abstract

How to obtain personalized quality of cloud/IoT services and assist users selecting the appropriate service has become a hot issue with the explosion of services on the Internet. Collaborative QoS prediction is proposed to address this issue by borrowing ideas from recommender systems. However, there is still a challenging problem as how to incorporate contextual factors into existing algorithms to realize context-aware QoS prediction as contextual factors play a crucial role in QoS assessment. In this paper, we propose a general context-sensitive matrix-factorization approach (CSMF) to make collaborative QoS prediction. By considering the complexity of service invocations, CSMF models the interactions of users-to-services and environment-to-environment simultaneously, and make full use of implicit and explicit contextual factors in the QoS data. Experimental results show that CSMF significantly outperforms the-state-of-art methods in metric of prediction accuracy. Particularly, when the QoS data is very sparse, CSMF is more effective and robust.

Introduction

With the rapid development and widespread deployment of cloud computing and Internet of Things (IoT) technologies, more and more homogeneous services emerge on the Internet [1]. Specially, the amount of cloud-based IoT services will probably explode with the integration of Cloud, IoT and Service-Oriented Architecture [2]. QoS-based service evaluation has become more important [3], [4], as users can take a decision on choosing appropriate services with distinguishable quality values of candidate services. Recently, researchers have proposed collaborative QoS prediction approaches to solve this problem by drawing lessons from the recommender systems [5], [6], [7]. Assume there exist some cyber–physical systems (which are being linked to versatile individuals in physical space and social space [8]) in which the QoS data, user behavioral data and various contextual information can be collected. Based on the principle that similar users (or services) tend to observe similar quality scores on the same service (or user), collaborative QoS forecasting models are built to predict the unknown quality values given active users and services as well as associated contexts (Fig. 1 shows an example in which the observed QoS values are used to predict the unknown ones). This treatment exploits crowd intelligence to aid the QoS assessment and avoid intuitive data measurement, thereby save time and economic costs for both service providers and users [9]. Consequently, collaborative QoS prediction has become an increasingly active and highly problem-rich research area, and many efforts have been done based on either neighborhood-based collaborative filtering (CF) or matrix-factorization [9].

Nevertheless, in most of the existing works, only the feedback matrix which contains explicit quality records observed by users on different services are used for training and prediction. There is lack of consideration on exploiting contexts in the process of prediction. Context-awareness is the potential demand of the recommender systems [10], especially has been shown great application value in precision marketing of E-commerce. Naturally, it is valuable as for the task of QoS prediction of cloud/IoT services. On one hand, as all elements in IoT-aided cyber–physical systems are individualized and privatized [11], it is necessary to provide users with personalized predicting results. Therefore we need to consider the contexts of service invocation and detect the changes of quality of user-end to service-end in different contextual conditions. On the other hand, various sensors allow us to perceive rich contextual information about location, time, network condition and device configuration, under the environment of cyber–physical systems [11]. Utilizing these information can alleviate the inherent data-sparsity problem with the user-service QoS matrix, and thus improve the accuracy of QoS predictions. Clearly, with the deployment of large-scale cloud-based IoT service systems, context-aware QoS prediction will become increasingly influential. In this regard, there have been some exploratory works. For example, modifying the similarity measure in CF-based prediction model to use the location of services or users [12], [13], [14], relying on locations of users/services to regularize or adjust MF-based prediction model [15], [16], factorizing user-service matrix using tensor factorization with time information [17], or use a contextual condition to prefilter and aggregate the training records [18]. However, these methods can only use the contextual factors in one dimension, and this will suffer from a unsatisfactory performance. There continues to be a challenging problem as how to incorporating rich contextual factors into existing algorithm, due to their inherent limits of scalability and flexibility.

Motivated by these, we are expected to develop a general method for the task of context-aware QoS prediction. By considering the complexity of service invocations, a general context-sensitive matrix-factorization approach (CSMF) is proposed to model the interactions of users-to-services and environment-to-environment simultaneously. By this, CSMF can make full use of implicit and explicit contextual factors in the QoS data during the training process and further generate better prediction results given the active users, the target services and the associated contexts.

The remainder of the paper is structured as follows. We review some existing works that are most relevant to ours in Section 2. We give details of the proposed CSMF model and present the formal foundation in Section 3. Meanwhile, how to optimize the parameters of the CSMF model is presented. We measure the effectiveness of proposed methods via a set of experiments on real QoS data in Section 4 and end in Section 5.

Section snippets

Related works

Collaborative QoS prediction borrows the idea from either neighborhood-based collaborative filtering (CF)or matrix-factorization. Both methods have their advantages and disadvantages.

Shao et al. [5] at first proposed the use of neighborhood-based prediction model. Given a user u and a service i, the set of n-nearest neighbors of user u, Nun, is firstly determined using a similarity measurement on the user-service matrix of QoS data. Then historical quality values of service i offered by the top-

Problem formulation

Context-aware QoS prediction addresses modeling and predicting the quality of services by incorporating contextual factors into the prediction process except the user-service QoS matrix. Quality values are modeled as the function of not only services and users, but also the contexts. We formally define the predicting process as a function: User×Service×ContextQuality where User, Service, Quality are the domains of users, services and QoS, respectively, and Context specifies the contextual

Datasets

To evaluate the QoS prediction performance, we use WS-DREAM dataset,1 a large-scale dataset collected and maintained by Zheng et al. [3], [34]. The dataset consists of a total of 1 974 675 real-world web service invocation results are collected from 339 users on 5825 real-world web services via PlanetLab platform,2 which is a global research

Conclusion and future works

We have proposed a MF-based approach for making context-aware QoS-prediction of cloud services. The framework offers an efficient global optimization scheme that enables robust and accurate prediction results. The intensive experimental analysis indicates the effectiveness of our approach.

The principal contribution of this paper lies in two-folds. First, we suggest a context-aware prediction model for collaborative QoS evaluation. Second, we prove that exploiting contextual information in

Acknowledgments

This work is supported by the National Natural Science Foundation of China (61562090, 61472345, 61562092, 61402398), China Postdoctoral Science Foundation funded project (2016M592721), the Applied Basic Research Project of Yunnan Province (2014FA023, 2013FZ107), the Special Funds for Middle-aged and Young Core Instructor Training Program of Yunnan University and Program for Excellent Young Talents of Yunnan University (WX173602). The authors are grateful to reviewers for their constructive

Hao Wu received the bachelor’s degree in computer science from Zhengzhou University, in 2001, master and Ph.D. degrees in computer science from Huazhong University of Science and Technology in 2004 and 2007, respectively. He is an associate professor at School of Information Science and Engineering, Yunnan University, China. He has published more than fifty papers in peer-reviewed international journals and conferences. His research interests include service computing, information filtering and

References (40)

  • KongX. et al.

    Urban traffic congestion estimation and prediction based on floating car trajectory data

    Future Generation Comp. Syst.

    (2016)
  • ZhengZ. et al.

    Collaborative web service QoS prediction via neighborhood integrated matrix factorization

    IEEE Trans. Comput.

    (2013)
  • ShaoL. et al.

    Personalized QoS prediction for web services via collaborative filtering

  • ZhengZ. et al.

    WSRec: a collaborative filtering based web service recommender system

  • ZhengZ. et al.

    Collaborative reliability prediction of service-oriented systems

  • WuH. et al.

    Deviation-based neighborhood model for context-aware QoS prediction of cloud and iot services

    Future Gener. Comput. Syst.

    (2016)
  • AdomaviciusG. et al.

    Context-aware recommender systems

  • ChenX. et al.

    Web service recommendation via exploiting location and QoS Information

    IEEE Trans. Parallel Distrib. Syst.

    (2014)
  • WuJ. et al.

    Predicting quality of service for selection by neighborhood-based collaborative filtering

    IEEE T. Systems, Man, and Cybernetics: Systems

    (2013)
  • TangM. et al.

    Cloud service QoS prediction via exploiting collaborative filtering and location-based data smoothing

    Concurrency and Computation: Practice and Experience

    (2015)
  • Cited by (105)

    • AI-driven, QoS prediction for V2X communications in beyond 5G systems

      2022, Computer Networks
      Citation Excerpt :

      Finally, they propose two models to produce the final QoS prediction results from user side and service side respectively, based on a real-world dataset that verifies the effectiveness of their method. Due to the high impact of cloud computing in the field of scientific and business technology domains, QoS Prediction has drawn the attention of researchers in this domain as well [28–33]. In [28] Li et al. collect real cloud computing environment data, obtain the correlation between the hardware/software resource data and QoS attributes of the Cloud service and propose a novel QoS prediction approach based on Bayesian Network Model (BNM).

    • Accelerated Structured Matrix Factorization

      2024, Journal of Computational and Graphical Statistics
    • Convolutional Neural Network Based QoS Prediction with Dimensional Correlation

      2024, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
    View all citing articles on Scopus

    Hao Wu received the bachelor’s degree in computer science from Zhengzhou University, in 2001, master and Ph.D. degrees in computer science from Huazhong University of Science and Technology in 2004 and 2007, respectively. He is an associate professor at School of Information Science and Engineering, Yunnan University, China. He has published more than fifty papers in peer-reviewed international journals and conferences. His research interests include service computing, information filtering and recommender systems.

    Kun Yue received the B.Sc., M.Sc., and Ph.D. degrees in computer science from Yunnan University (Kunming, China), Fudan University (Shanghai, China) and Yunnan University (Kunming, China) in 2001, 2004 and 2009, respectively. He is currently a professor at Yunnan University, Kunming, China. His current research interests mainly include massive data analysis and uncertainty in artificial intelligence.

    Bo Li received the B.Sc. and M.Sc. from Yunnan University, Ph.D. degrees from Huazhong University of Science and Technology. He is an associate professor at School of Information Science and Engineering, Yunnan University, China. He has published more than forty papers in journals and conferences. His research interests include cloud computing, mobile computing and grid computing.

    Binbin Zhang was born in 1982. She received the Ph.D. degree in computer science from Peking University in 2011. She is currently a lecturer at Yunnan University. Her main research interests include virtualization and cloud computing.

    Ching-Hsien (Robert) Hsu Ching-Hsien Hsu is a professor and the chairman in the CSIE department at Chung Hua University, Taiwan; He was distinguished chair professor at Tianjin University of Technology, China, during 2012–2016. His research includes high performance computing, cloud computing, parallel and distributed systems, big data analytics, ubiquitous/pervasive computing and intelligence. He has published 100 papers in top journals such as IEEE TPDS, IEEE TSC, IEEE TCC, IEEE TETC, IEEE System, IEEE Network, ACM TOMM and book chapters in these areas. Dr. Hsu is serving as editorial board for a number of prestigious journals, including IEEE TSC, IEEE TCC. He has been acting as an author/co-author or an editor/co-editor of 10 books from Elsevier, Springer, IGI Global, World Scientific and McGraw-Hill. Dr. Hsu was awarded nine times distinguished award for excellence in research from Chung Hua University. He is vice chair of IEEE TCCLD, executive committee of IEEE TCSC, Taiwan Association of Cloud Computing and an IEEE senior member.

    View full text