A Service-Oriented Approach for the i* Framework
Abstract
New application areas such as e-Business, application service provision and peer-to-peer computing all call for very complex software systems which effectively support “on-line” enterprise processes. To build such systems, practicing software engineers are discovering the effectiveness of using organizational modeling techniques to facilitate the elicitation of requirements for information systems and also for guiding and supporting the software production process.
In this context, the i* Framework is one of the most well-founded organizational modeling techniques in use today. It mainly focuses on: a) the representation of social and intentional relationships among the network of actors of an enterprise, and b) the representation of the internal behaviors required to satisfy actor dependencies. The i* framework supports the description of organizational networks made up of social actors that have freedom of action, but that depend on other actors to achieve their objectives and goals. 
Despite the well-known advantages of the i* modeling approach, there are certain issues that still need to be improved to assure their effectiveness in practice. In order to accurately identify areas of strength as well as weaknesses of i* in real case studies, empirical evaluations of this framework must be conducted in practice. 
One of the objectives of this thesis is to present an empirical evaluation that enables us to identify and to understand what the practical problems of i* are. We present the lessons learned, both in terms of the strengths of i* and in terms of the detected weak points that need to be overcome. Solutions for these weak points are also proposed as an initial response to the results of the empirical evaluation. 
We consider that service-orientation is currently considers as a very promising paradigm to deal with the complexity of modeling the IT systems. In this sense, the main objective of the thesis is to define a service-oriented architecture to solve the problem of i* complexity in real-life cases. The proposed architecture distinguishes three abstractions levels (services, process and protocols) and describes a methodological approach to align the business models produces at these abstraction levels.
Our service-oriented approach considers the following aspects: a) A conceptual modeling language, based on the i* primitives, which defines the modeling concepts and their corresponding relationships. b) A service-oriented architecture specific for the i* framework that define the service components and the modeling diagrams. c) A business modeling method to represent services at the organizational level.
The business services and the service components have been precisely defined in terms of properties and relationships. It is important to point out that services components have been designed based on the intended use of this abstraction in representing services at the organizational level. 
With the extensions proposed in this thesis, our intention is to overcome the current limitations that practitioners face when using i* in its current state. In fact, these extensions are intended to both, solve the problems that were detected, and to make the practical application of the method easier.