Reputation Management for Composite Services in Service-Oriented Systems

Introduction

With the help of Web Services standards and technologies, development and deployment of services on the Web has become easier, and is no longer limited to large enterprises. A large number of individual entrepreneurs as well as small and medium scale enterprises have started deploying their services on the Web (Saulles, 2008). The ultimate goal of the Web services technology is enabling the use of Web services as independent components in “service-oriented enterprises” that are automatically (i.e., without human intervention) formed as a result of consumer demand and which may dissolve post demand-completion. The service-oriented Web (Service Web) thus represents an attractive paradigm for tomorrow’s interactions spanning a wide range of domains from e-economy to e-science and e-government. The number of Web users has also grown phenomenally from 0.4 percent of world population in 1995 to 23.3 percent in December 2008 (Internet World Stats, 2008). A significant portion of these users have started consuming services over the Web. Consumers can now access a large pool of services with varying qualities providing similar kinds of functionalities (Hull, Benedikt, Christophides, & Su, 2003). For example, a traveller can access a large number of services offered by online travel agencies (providing similar functionalities). A consumer thus faces a difficult task of choosing the best service(s) that meets his/her requirements, as some services may not even deliver what they promise (resulting in bad track records). This implies that a service consumer needs to trust the ability of the provider to deliver the required functionality before starting the interaction. Hence, trust is one of the essential and important criteria a consumer can use in selecting the right services. The importance of trust management in Web Services is evident from the interest in both the academic and business communities (Ruohomaa & Kutvonen, 2005; Zhao & Varadharajan, 2008).

A wide range of approaches for managing trust on the Service Web have been proposed in the literature. These range from enforcing rules and regulations to the use of trusted third party certificates. On one hand, rules and regulations are not globally enforceable, while on the other, third party based trust is not scalable to the Service Web. Security technologies based approaches have also been used in some application scenarios (Anderson et al., 2005; Lin, Varadharajan, Wang, & Pruthi, 2005) to establish trust. These approaches are mainly successful in achieving authorization, authentication and confidentiality, but not necessarily trust. The lack of applicability of previous trust management approaches is due to the following reasons: (i) the Service Web is an open and dynamic environment, (ii) the service providers may not be known a priori to the customers and vice versa, and (iii) a central authority for trust management is not feasible. Consequently, reputation based approaches have been proposed as a practical alternative to manage trust on the Service Web (Ruohomaa, Kutvonen, & Koutrouli, 2007; Chang & Hussain, 2006), in both industry applications and academic research (Resnick, Kuwabara, Zeckhauser, & Friedman, 2000; Tennenholtz, 2004).