A Collaborative Ranking Approach for Discovery and Selection of Cloud Services

Introduction

Cloud computing is gaining dominance over the last few years along with its various features such as, on-demand ease of access of computing resources as well as software services. Computing facilities are designed and utilized as a service using virtualization techniques and used as automated business logic in both public and private sectors. Along with various benefits data privacy, data confidentiality and trust establishment are considered to be the main security concerns for an organization to move its data to the cloud platform (Xu et al., 2015; Elfirdoussi, Jarir & Quafafou, 2012). The hasty developments of cloud computing means that cloud services have become the main computing mode on the Internet. Numerous services are offered over cloud to offer analogous functionalities. Despite the fact, the difficulty in identifying trustworthy services has fascinated the attention of researchers (Xu et al., 2015; Elfirdoussi, Jarir & Quafafou, 2012). For this reason, the concepts of trust and reputation had been brought to assess the reliability of cloud services (Li, Zhou, & Yang, 2012).

Reputation is a prejudiced assessment of a cloud service based on personal experience of individual or the advocacy of other users. In recent times, a variety of reputation systems have been proposed to deal with the challenges posed by open and dynamic cloud service environments (Itani, Ghali, Kayssi, & Chehab, 2014). The focus of majority of these systems is on computation of reputation ratings, reputation management, experience, and other features of dynamic environments and provides an appropriate solution to users (Malik & Bouguettaya, 2009; Rathore & Suman, 2013a, 2013b,2013c, 2016; Yuan, An, & Wang, 2009; Trang, Zhao, & Yang, 2010). However, the survival of biased ratings affects the accuracy of trust evaluations to a great extent. At present, the focus of these reputation models are mainly on the accuracy of trust evaluations (Wang, Zheng, Sun, Zou, & Yang, 2011; Malik & Bouguettaya, 2009a,2009b); however, these existing methods are limited by personality preference.

At present, to deal with the issue of trustworthiness, computation of the reputation is one of popular method (Elfirdoussi, Jarir & Quafafou, 2012). Reputation is the subject of a lot of interdisciplinary research and can be defined as “the collected and processed information about an individuals' behavior as observed by others” (Wang, Chao, Lo, Lin, & Wang, 2011). The user can trust the service to a certain extent, based on aggregated feedback given by earlier consumers can be considered as the reputation score of a cloud service. A common attribute in existing cloud computing models is that a universal reputation score is designed through these reputation systems and the value is the same for all consumers. It is commonly used in profit-making location, such as eBay's feedback mechanism or Epinions (Li, & Wang, 2011). However, in general the reputation is potentially subjective. Feedback ratings are based on the individual’s usage experience of a cloud service, which always needs to be seen in the context of the users experience and expectations of a service.

Consider for example two cloud service consumers A (denoted as CA) and B (denoted as CB). They are both looking for a cloud storage service. An industry expert has extremely precise requirements for confidentiality. CB is an amateur whose main aim is to obtain cheap storage. Now consider CB a free storage service is expected to be extremely glad as the cost is minimal and if usability is ok he/she will rank this service very highly. CA might read the small print regarding data policies and is likely to find issues that are not to her liking; so they would rank the service somewhat lower. Of course both are correct in their individual rankings. A global reputation score would not reflect the service users accurately.