A two-stage location-sensitive and user preference-aware recommendation system
Introduction
A cloud recommendation system is a middleware which provides the desired cloud services for cloud customers using managing multiple cloud services and negotiation between the cloud providers and cloud consumers. In a cloud environment, the recommendation system manages uncertainty situation in cloud services, heterogeneity, and dimensionality. It can play a significant role in helping cloud customers to select the most appropriate cloud services with lower cost and more high quality (Chauhan et al., 2019, Elhabbash et al., 2019, Tordsson et al., 2012).
Based on pay-per-use logic in the cloud computing model, each user can use on-demand cloud services and just pay cost considering the amount of own consumption (Bhat et al., 2011, Weinman, 2016). In recent years, a drastic competitive between cloud providers have been constructed. Providers in different countries produce cloud services which are functionally equivalent. Consequently, predicting the value of quality of service (QoS) and selecting the most proper service among candidate services is a difficult task, since the quality of service highly depends on various factors such as the location of providers and users, the clarity in precise expressing service requirements, and etc. (Bushehrian and Mohammadi, 2016, Esposito et al., 2015a). Hitherto, a large number of methods for identifying and selecting the suitable cloud services have been presented (Ding et al., 2017, Esposito et al., 2015a, Sun et al., 2016, ur Rehman et al., 2011). However, the presented methods cannot support immature customers and cannot satisfy their demands.
It can be argued that the following factors in the cloud service recommendation field need special attention because these factors can significantly affect the quality of cloud services, the satisfaction of cloud customers, the reputation of providers, satisfying service level agreement (SLA) and so on.
- –
The lack of specifying service requirements in precise terms: The use of cloud services is being grown by users to satisfy their requirements. Some cloud users are immature and inexperienced and cannot express their needs accurately and numerically. Hence, they use linguistic expressions (such as medium, high, low and so on) to express their needs. For example, the inexperienced users to use cloud infrastructure (IaaS) and to meet computing requirements may express their need as Hard Disk (HDD) = Medium, RAM = High, computing capacity = High, network bandwidth = Medium. Without a flexible recommendation system, such requests cannot be met.
- –
Prevention of service level agreements (SLA) violation: In a dynamic cloud environment, contracts between customers and providers should be properly managed so that be prevented SLA violations. To arrive this goal, the proposed recommendation system tunes different fuzzy weights for each QoS so that leads to the prevention of SLA violation. Indeed, the proposed method uses a fuzzy value tuned by the neuro-fuzzy network instead utilization of a crisp value for a QoS metric.
- –
Impacts of the geographical distribution of providers and customers on QoS: Some hidden factors such as the geographical location of cloud service customers or providers and the distance between them can easily affect QoS values. Hence, knowledge of the geographical location of customers and providers has a significant effect on recommendation system decision.
- –
Differences in country laws regarding the use of cloud services: Since, the different providers and customers have geo-distribution, ensuring the privacy and security of cloud customer data is a major challenge, especially when data is located in other countries. Because different countries usually have different laws. Hence, guaranteeing confidentiality and security on data is difficult. Moreover, concerning legal borders in some situations, there is a restrict restriction to use cloud services out of country borders (El-Gazzar & Wahid, 2015). For example, considering the European Union (EU) decree, all personal data which are produced within the EU should not leave the EU. Unless the destination country has sufficient credibility to protect the privacy of the information (El-Gazzar and Wahid, 2015, Essén et al., 2018). Consequently, considering the location of providers and consumers and laws of different countries are essential to select cloud service. In this paper, the proposed method can find the closest services to any cloud customer and can protect the security and confidentiality of information a large extent. To address the mentioned issue, this paper proposes a new technique using an innovative neuro-fuzzy method which combines the advantages of the fuzzy logic with the generalizability and the prediction power of neural networks. The neuro-fuzzy method analyzes the specifications and requirements announced by the customer and can find the closest services to the customer and estimate their price.
In this paper, to more clarity, the proposed method has been presented to provide IaaS services so that the cloud customer can request own desired storage as two ways, numerical for example Hard Disk (HDD)= 13TB or as a linguistic term, for example, Hard Disk (HDD) = Medium. The fuzzy logic has been used to construct a recommendation system which can handle all types of customer requests i.e. numerical and linguistic terms. To better estimate the QoS and select a suitable service by the recommendation system, we merge fuzzy logic with neural network science. When a customer sends a cloud service request to a recommendation system, the recommendation system takes the service price and other quality characteristics from providers and considering the distance between the provider and customer and the preferences of customer such as the quality and price of services recommends a suitable service to the customer. In this method, after requesting a specific service by the customer, the recommendation system creates a sorted list of each feature of service according to the services announced by the providers. After that, the recommendation system defines a range of values for the features and divides the range of values into sections and denominates each section with a linguistic term such as high, low, medium.
Fig. 1 illustrates a fuzzy range for HDD instance which has been divided into three sections namely Low, Medium, and High. It is worth noting that, the recommendation system uses an iterative neuro-fuzzy learning algorithm to tune the fuzzy range of feature values during the decision making to select cloud services. In the last step of the decision-making process, the recommendation system converts its fuzzy decision to a crisp decision by applying the Defuzzification method. The function of the proposed method has been explained in Section 4 in more detail. The presented method can attain high accuracy in the selection of the most appropriate neighbor cloud services for customers so that, the response time be minimal and the offered cloud service has most adaptation with the demands of cloud customers. It is necessary to mention, the proposed method is suitable for all types of cloud customers (experienced and inexperienced users). Inexperienced customers can linguistically express their requirements and attain a proper service.
Hitherto, various techniques have been presented in the area of cloud service selection to identify and recommend the most suitable cloud service to customers (He et al., 2014, Jin et al., 2019, Luo et al., 2016a, Peng et al., 2018, Ren and Wang, 2018, Ryu et al., 2018a, Tang et al., 2016a, Yin et al., 2019, Yin et al., 2017, Zheng et al., 2009). However, existing cloud service selection methods have not considered the immature customers. In this paper, the mentioned challenge has been solved by combining the various powerful science i.e. the fuzzy logic science and neural-network science. In summary, the main contributions of paper are as follows:
- –
A Location clustering technique has been proposed based on a fuzzy clustering algorithm (FCA) where the users can belong to two or more regions, and geographical locations are expressed in the form of latitude–longitude. A DoP (the degree of proximity) metric has been introduced where denotes the membership value of data point i (customer i) to cluster j (provider j). Using the FCA, the recommendation system can find providers close to the customers who are placed in a geolocation zone in the earth map.
- –
An iterative adaptive neural fuzzy recommendation algorithm (IANFRA) has been proposed that can usefully manage the needs expressed verbally and numerically (a crisp value) and predict features and price of services based on the desired preferences of customers to recommend a suitable service.
- –
Wide experiments have been conducted to examine the performance of the proposed method using two dataset. The first dataset is a large set of real web services called WS-DREAM (Zheng & Lyu, 2008) and the second dataset is an expanded dataset of cloud IaaS services which offered by https://www.digitalocean.com/ (2021).
The remainder of the paper is organized as follows. In Section 2, the basic concepts related to proposed method are introduced. In Section 3, the related works are reviewed and an overview of the proposed IANFRA is presented in Section 4. The experiments has been explained in Section 5 in detail and the evaluation of the presented recommendation system has been done in Section 6, and finally the conclusion is presented in Section 7.
Section snippets
Background
To describe the IANFRA, some basic definitions should be introduced for the concepts used in this article. These definitions cover two areas including the fuzzy clustering algorithm and the iterative adaptive neural fuzzy recommendation algorithm.
Related works
In this section, the related works in the area of cloud service selection and recommendation have been reviewed. Various studies have been conducted in the field of recommendation of cloud services to customers, nevertheless, the presented works have assumed that all cloud customers can express their own demands as exact and numerical. Therefore, an immature customer cannot receive proper services from these recommenders. According to the latest studies, existing approaches to provide a
The proposed recommendation system
As previously stated, in the field of cloud service selection and recommendation, most of the available studies presume all customers are expert and express their requirements, such as the amount of the hard disk (HDD) or RAM required for an IaaS service, in an explicit and precise numeric way (Chen and Ma, 2015, Chen et al., 2015, Jin et al., 2019, Kuang et al., 2018, Luo et al., 2016a, Ren and Wang, 2018, Ryu et al., 2018a, Shao et al., 2007, Tang et al., 2016a, Yin et al., 2019, Yin et al.,
Experiments
In this section, we have conducted comprehensive experiments to demonstrate the ability of the proposed recommendation system to identify and recommend appropriate services according to the location and preferences of users. All the experiments have been implemented in MATLAB R2019b on Windows10 64 bit OS with Intel Core i7-8750 2.20 GHz processor and 16 GB RAM. As mentioned stated, one of the primary advantages of the proposed recommender is satisfying the demands of all types of users
Evaluation
The purpose of evaluation is to answer the following questions:
(1) how is the performance of FCA compared with other popular clustering methods such as k-means algorithm and subtractive clustering algorithm? (2) How generalizable and scalable is the proposed method in large datasets? (3) What is the advantage of the proposed method compared with other related methods? and how is the performance of IANFRA compared with other methods?
Conclusion
In this paper a novel method has been proposed to identify and to recommend suitable cloud services considering the demands of the cloud users and the geolocation distribution of them. The proposed method uses a fuzzy clustering algorithm to find providers close to the users. Then, an iterative adaptive neural-fuzzy recommendation algorithm (IANFRA) is represented to identify and to recommend suitable services considering the preferences of the user. The immature users can easily express their
CRediT authorship contribution statement
Neda Mohammadi: Conceptualization, Investigation, Formal analysis, Validation, Visualization, Methodology, Writing – original draft, Writing. Abbas Rasoolzadegan: Conceptualization, Project administration, Supervision, Writing – review & editing.
Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
References (86)
- et al.
Analysis of k-means and k-medoids algorithm for big data
Procedia Computer Science
(2016) - et al.
Brokering in interconnected cloud computing environments: A survey
Journal of Parallel and Distributed Computing
(2019) - et al.
Constructing membership functions using interpolation and measurement theory
Fuzzy Sets and Systems
(1995) - et al.
Exploiting web service geographical neighborhood for collaborative qos prediction
Future Generation Computer Systems
(2017) - et al.
Your neighbors are misunderstood: On modeling accurate similarity driven by data range to collaborative web service QoS prediction
Future Generation Computer Systems
(2019) - et al.
Image segmentation using K-means clustering algorithm and subtractive clustering algorithm
Procedia Computer Science
(2015) - et al.
Utilizing customer satisfaction in ranking prediction for personalized cloud service selection
Decision Support Systems
(2017) - et al.
Patient access to electronic health records: differences across ten countries
Health Policy and Technology
(2018) - et al.
Another look at measures of forecast accuracy
International Journal of Forecasting
(2006) - et al.
Social collaborative service recommendation approach based on user’s trust and domain-specific expertise
Future Generation Computer Systems
(2018)
A multi-criteria collaborative filtering recommender system for the tourism domain using expectation maximization (EM) and PCA–ANFIS
Electronic Commerce Research and Applications
Application of subtractive clustering for power transformer fault diagnostics
Procedia Engineering
An SVM-based collaborative filtering approach for top-n web services recommendation
Future Generation Computer Systems
TAP: A personalized trust-aware QoS prediction approach for web service recommendation
Knowledge-Based Systems
Cloud-FuSeR: Fuzzy ontology and MCDM based cloud service selection
Future Generation Computer Systems
Time series forecasting for dynamic quality of web services: an empirical study
Journal of Systems and Software
Cloud brokering mechanisms for optimized placement of virtual machines across multiple providers
Future Generation Computer Systems
Deviation-based neighborhood model for context-aware QoS prediction of cloud and IoT services
Future Generation Computer Systems
Collaborative QoS prediction with context-sensitive matrix factorization
Future Generation Computer Systems
Context-aware QoS prediction for web service recommendation and selection
Expert Systems with Applications
Exploring latent features for memory-based QoS prediction in cloud computing
Spatio-temporal context-aware collaborative QoS prediction
Future Generation Computer Systems
Toward the next generation of recommender systems: A survey of the state-of-the-art and possible extensions
IEEE Transactions on Knowledge and Data Engineering
Price modeling of iaas providers-an approach focused on enterprise application integration
FCM: The fuzzy c-means clustering algorithm computer & geosciences
Volume
Cloud computing: A solution to geographical information systems(GIS)
International Journal on Computer Science and Engineering
Association rule mining for failure recovery strategy selection in composite web services
Journal of Telecommunication, Electronic and Computer Engineering (JTEC)
Hybrid collaborative filtering algorithm for bidirectional web service recommendation
Knowledge and Information Systems
A hybrid approach to web service recommendation based on qos-aware rating and ranking
A ranking-oriented hybrid approach to QoS-aware web service recommendation
Predicting quality of service via leveraging location information
Complexity
Service models and pricing schemes for cloud computing
Cluster computing
Evolution-fuzzy rule based system with parameterized consequences
International Journal of Applied Mathematics and Computer Science
Compatibility-aware cloud service composition under fuzzy preferences of users
IEEE Transactions on Cloud Computing
Music recommendation and query-by-content using self-organizing maps
Location-based social network recommendations with computational intelligence-based similarity computation and user check-in behavior
Concurrency Computations: Practice and Experience
Cloud brokerage: A systematic survey
ACM Computing Surveys
Smart cloud storage service selection based on fuzzy logic, theory of evidence and game theory
IEEE Transactions on Computers
Smart cloud storage service selection based on fuzzy logic, theory of evidence and game theory
IEEE Transactions on Computers
Indirect estimation of deformation modulus of an in situ rock mass: an ANFIS model based on grid partitioning, fuzzy c-means clustering and subtractive clustering
Geosciences Journal
Optimization of a truck-drone in tandem delivery network using k-means and genetic algorithm
Journal of Industrial Engineering and Management (JIEM)
A survey on web service QoS prediction methods
IEEE Transactions on Services Computing
Cited by (7)
Long tail service recommendation based on cross-view and contrastive learning
2024, Expert Systems with ApplicationsA heterogeneous E-commerce user alignment model based on data enhancement and data representation
2023, Expert Systems with ApplicationsAn optimized recommendation framework exploiting textual review based opinion mining for generating pleasantly surprising, novel yet relevant recommendations
2022, Pattern Recognition LettersCitation Excerpt :A user may lose interest in an RS if RS remains restricted to relevant or popular items and exclusively focuses on accuracy [4]. Instead, the RS should include a diverse, novel and unexpected set of preferences [5,6]. Diversity and novelty-based RS [7] avoid the overrated and similar recommendations.
Application of Recommendation Algorithms Based on Social Relationships and Behavioral Characteristics in Music Online Teaching
2024, International Journal of Web-Based Learning and Teaching TechnologiesMeasuring Uncertainty to Extract Fuzzy Membership Functions in Recommender Systems
2023, Journal of Computer ScienceNews Recommendation with Multi-views Emotion Analysis
2022, ACM International Conference Proceeding Series
The code (and data) in this article has been certified as Reproducible by Code Ocean: (https://codeocean.com/). More information on the Reproducibility Badge Initiative is available at https://www.elsevier.com/physical-sciences-and-engineering/computer-science/journals.