Interface inheritance for object-oriented service composition based on model driven configuration Vincenzo D’Andrea Ioannis Fikouras Marco Aiello DIT, Univ. of Trento BIBA, Univ. of Bremen DIT, Univ. of Trento Via Sommarive, 14 Hochschulring 20 Via Sommarive, 14 38100 Trento 28359 Bremen 38100 Trento Italy Germany Italy [email protected] [email protected] [email protected] ABSTRACT the pursuit of differentiation through variety leads to unique In today eCommerce environments, customers have to deal products and services [16, 19]. This strategy is known as with a wide variety of alternatives, both in terms of service “mass customization”. Mass customization is defined as offerings as well as service providers. They risk to be over- “when the same large number of customers can be reached whelmed by the complexity of alternatives, thus reducing as in mass markets of the industrial economy, and simul- the usefulness of the experience and consequently the like- taneously they can be treated individually as in the cus- lihood of transactions. There is an increasing need for new tomized markets of pre-industrial economies” [12]. Accord- ways to reduce the perceived complexity. Service-oriented ing to [31] the objective of mass customization is “to deliver computing can help the user cope with this problem. With goods and services that meet individual customers needs services, interfaces no longer hide units of code, but provide with near mass production efficiency”. Online transactions access to complex functionality equivalent to that of entire and specifically eCommerce environments differ greatly from conventional applications. conventional commercial transactions. Online transactions achieve greater execution speeds and can bridge greater dis- We introduce a methodology for extended service compo- tances than traditional commerce. Furthermore purely dig- sition derived from model-driven configuration and object- ital products (i.e., information services or digitized media) oriented systems. By focusing on the concept of interfaces, can be discovered, adapted, evaluated, purchased, paid for and applying it to the object-oriented concept of inheri- and delivered by a single service platform within a very short tance, we propose an innovative approach to composition timeframe at any time of day or place on earth [32]. More- that takes into account how the composed services can be over such platforms compared to conventional sales facilities recognized or accessed via the composing service. In or- (i.e., brick and mortar stores) are quick and cheap to imple- der to set the stage, we discuss the similarities between ment as well as adapt to new requirements even in not previ- Service Oriented Computing, Object-Oriented Configura- ously predetermined ways [32]. This allows for the rapid and tion and Object-Orientation. In addition, we provide an inexpensive deployment of on-line stores offering advanced overview of knowledge-based systems, described as software functionality (such as rearranging the product palette for systems built by capturing the knowledge used by experts, individual customers) impossible to implement in brick-and- and more specifically object oriented configuration for im- mortar facilities. plementing service composition. On the other hand, customers in an eCommerce environ- Categories and Subject Descriptors ment are faced with more information, resulting from a H.1 [Information Systems]: Models and Principles; D.1.5 wider variety of alternatives both in terms of service offer- [Software]: Programming techniques—Object-oriented Pro- ings as well as service providers. However the processing of gramming this information occurs based on the same knowledge and information processing capacity available to the customer as in conventional shopping scenarios [32]. These constraints, General Terms unaffected by new technologies, result in a significant draw- Web services, Object-oriented programming, Model driven back to high variety strategies. A customer overwhelmed configuration by the amount of available products or frustrated by their complexity is less likely to complete the transaction and pur- 1. INTRODUCTION chase the product, and more likely to delay the decision or “It is the customer who determines what a business is” [13] leave the shop altogether [18]. This behaviour illustrates by attempting to address specific needs and express his per- the need for new ways for retailers to reduce the perceived sonality through custom-made products and services [26]. complexity of their products. Advanced functionalities are Customers thus drive vendors to strive for product palettes designed to help the user cope with a large amount and at with an ever increasing number of variants. Consequently the same time a significant complexity of product data. The advance functionality necessary to accomplish the vision of mass customization may be offered by service composition functionality implemented in a service-oriented infrastruc- In [14], we have shown how Model Driven Configuration ture. theory can be exploited for service composition and orches- tration, in [11] we have shown the analogies relating object- Service oriented computing (SOC) is a new computing para- oriented programming and service-oriented design. In this digm in which complex systems are built on the basis of ba- paper, we propose a methodology for extended service com- sic distributed autonomous services by abstracting on the position derived from model-driven configuration and object- actual implementation and location of the various services oriented systems, having the notion of service as the central [24]. This paradigm allows for a high distribution of the building block. By focusing on the concept of interfaces and workload, for the building of complex system yet dynami- applying it to the object-oriented concept of inheritance, we cally and easily scalable. Following the “Service Oriented propose an innovative approach to composition that takes Computing Manifesto” [25], SOC is more formally defined into account how the composed services can be recognized in terms of services, that is: or accessed via the composing service. We propose a classi- fication of service composition, derived from the concepts of Services are autonomous platform-independent inheritance, interface inheritance, and object composition. computational elements that can be described, For instance, from the notion of object composition we de- published, discovered, orchestrated and programmed rive the definition of Opaque Composition, that is, a service using XML artifacts for the purpose of develop- is composed by other services without informing the external ing massively distributed interoperable applica- world of the details of the composing services. tions. The paper is organized as follows. In Section 2 we provide an overview of knowledge based systems. The paper then pro- The best-known example of service-oriented technology is ceeds focusing on the use of knowledge-based construction that based on web services. In [10], web service are described systems, specifically object oriented configuration for imple- as menting service composition. In Section 3, we present a dis- cussion of the similarities between Service Oriented Comput- a networked application that is, able to interact ing, Object-Oriented Configuration and Object-Orientation, using standard application-to-application Web pro- in order to bridge the gap between model driven configura- tocols over well defined interfaces, and which is tion and services. Section 4 presents the main results of the described using a standard functional description paper, that is, a methodology for the composition of services language. based on object-oriented configuration. Concluding remarks and open issues are summarized in Section 5. In the SOC paradigm the emphasis shifts from the engineer- ing of appropriate isolated applications towards the integra- 2. KNOWLEDGE-BASED SYSTEMS tion, orchestration and choreography of a set of independent We focus on the use of knowledge-based construction sys- services over a network. Typical distributed systems prop- tems, specifically model-driven variant configuration for im- erties [8] become of paramount importance in this setting, plementing service composition. The following section gives most notably: heterogeneity, openness, security, scalability, thus a broad overview of knowledge-based systems. failure handling, concurrency, transparency. Furthermore, in the SOC model no fixed synchronous bindings are es- Knowledge-based systems are defined in [1] as: tablished, but rather the computational elements follow the find-bind-use model. computer programs which (a) use knowledge and If the scene is that of a web of autonomous computational inference procedures (b) to solve problems which, elements that offer simple services exposing their interfaces, if addressed by a human, would be regarded as then the challenge is that of creating massively distributed difficult enough to require significant expertise. applications offering added value by taking advantage of the basic services. In other words, service composition is the cornerstone for the success of the SOC vision. For the purposes of this paper we use the following definition of software systems built by capturing the knowledge used by Various approaches to service composition have been pro- experts and structuring it according to a specific method, in posed in the literature. On one extreme are those who con- order to solve problems requiring application domain