Requirements specification and analysis is a critical step in any system engineering process and the quality of identified requirements is crucial for the success of projects. A complete understanding of stakeholders' problems and their concise and clear representation as well as a representation understandable for stakeholders helps to design software that meet customers' needs. With the help of the participative card game CUTA we want to bridge the gap between stakeholder and developer. In several interviews with stakeholders from different domains we use our own adapted version of CUTA as an alternative approach to identify customized business processes. In this process our developed interview technique and our adapted CUTA was evaluated as well as transformed to formal models which provide a concise and clear representation for system developers.

In this paper, we describe our integrated approach to improve requirements elicitation and system specification by supporting a strong and direct involvement of all stakeholders, including non technical personnel, in the early phases of a project. This so-called CUTA4UML approach comprises a framework of methods, tools and feedback cycles that is based on the concept of Participatory Design (PD) and instantiated via an extended version of a user-driven "card game" (CUTA). This card game can be executed on paper or via electronic means. Our focus will be on the system's dynamic aspects which are, always, especially tricky to cover. We will also take a closer look at the inherently involved problem to match the rather informal results of such an card-based approach to a quite formal (UML -- Activity Diagrams) modeling technique -- a step necessary to make the results useful in a structured development process and, so far, one of the weak points in all of these approaches. All our concepts and conclusions are based on a sound experimental basis and have been evaluated and discussed with industrial partners, mostly in a context of small and medium sized companies.

The failure of projects is often based on difficulties in the direct and indirect communication between developer and customer teams. For the customer it is hard to understand the frequently used terms, process models, and technological concepts. Developers have a hard time to understand domain specific processes and structures, and exhibit a tendency to abstract concrete examples to higher level constructs. This makes it difficult for the customer team to understand the developing architecture and processes, and to match their needs and intentions, during evaluation phases, to seemingly unrelated generalised drawings and specifications. This set of problems is especially hard to avoid during the development of dynamic, distributed systems with multiple nodes and possibly asynchronous behaviour.Our research tries to mediate in this conflict by providing a customer-friendly reference model as basis of communication in early development phases. This model presents and simulates the dynamic aspects of (distributed) systems without immediate abstraction from examples and allows for a stepwise generalisation and evaluation with help of the customer team. In its final version the reference model will serve as a requirements statement for the professional developer.