Skip to content. | Skip to navigation

Informatik 5
Information Systems
Prof. Dr. M. Jarke
Sections
Personal tools
You are here: Home Staff Peter Haumer Abstract PhD Thesis - Peter Haumer

Abstract PhD Thesis - Peter Haumer

Requirements Engineering with Interrelated Conceptual Models and Real World Scenes

When creating a requirements document the objective is to establish a complete, consistent, and unambiguous description of intended changes for a given application domain. Descriptions of the application domain itself and the requirements are usually expressed using textual representations, but frequently also using conceptual models depicting, e.g., static properties or dynamic behaviour on an abstract and more formal level. The process which ultimately leads to these descriptions is called Requirement Engineering (referred to as RE in the following).

RE is also a cooperative learning process. Stakeholders from different backgrounds (having different experiences and objectives) and requirements analysts have to communicate with each other for eliciting and understanding requirements and models, as well as for detecting gaps and inconsistencies while validating them. The quality of conceptual models created during RE heavily depends on successful stakeholder involvement. This applies especially to the quality of current-state descriptions and models as well as change definitions on current-state models, because only the stakeholders dealing with the existing system in use (e.g. domain experts, system users, maintenance people etc.) have the fundamental knowledge about specific properties, behaviours, problems etc.

Nevertheless, involving stakeholders into the model creation and validation process is a difficult task. It is not enough to just bring stakeholders together in a meeting room and interview them about their work practice and requirements. Experiences show that stakeholders do not explicitly know what their requirements are, nor can they just explain what they are actually doing without a concrete frame of context to refer to. Moreover, many stakeholders, who potentially can provide valuable information about the problem domain, have not the ability to abstract and structure their knowledge nor to understand the abstractions made by the requirements engineer. However, stakeholders react to concrete instances and examples, especially when they come from their own familiar working environment. Thus, as argued for many techniques for ethnographical, contextual, and participatory design observing and talking to people while they are working in the field for understanding their work cannot be replaced by talking about it in meetings. Consequently, to be able to communicate these observations with others not being present to the observation they have to be captured; captured as rich as possible.

Scenarios are one possible form of captured working observations. In general, they represent concrete examples of current and future system usage. The use of textual scenarios, in addition to conceptual models, improve the quality of the RE process and descriptions produced, because scenarios in general are less abstract and therefore easier to develop and communicate, especially by those stakeholders with less formal training but important domain knowledge.

In our work we utilised — in addition to textual scenarios — richer representations for scenarios which we call real world scenes (recorded multimedia representations of system usage) which are widely used in the areas of video-supported participatory/user-centred design techniques, ethnography, and workplace culture for understanding existing systems and working environments, but also for the envisionment of future systems and different system changes. Experiences gained in these approaches show that the use of rich media in RE processes leads to a better understanding of the usage domain, enforces focused observation of (temporally and/or spatially) distributed aspects, avoids abstractions based on presumptions, enables repeatability of results and late reflections.

As a result, both types of representations, abstract and concrete ones, are needed interchangeably for RE. The requirements engineer needs to abstract for handling the complexity of phenomena, but in some situations it might be wrong to abstract to soon, because many important cases and variants of the abstracted phenomena might be unconsidered and lost. Stakeholders have not the ability to abstract; they need to refer to instances that are more concrete.

However, on the one hand, despite the advantages of using real world scenes, participatory design techniques do not support a tightly integrated management of real world scenes and conceptual models. Conceptual modelling techniques, on the other hand, do not offer a tight integration with scenarios and scenes. Our approach is motivated by the objective to provide this missing link: the integration of real world scenes and scenarios with conceptual descriptions.

In the following, we outline the basic solution idea for providing the missing integration in a concrete problem domain. Thus, in Figure 1 (left) we depict the typical way of working before the integration proposed by our approach. Analysis experts like the one depicted in the lower left corner produce different types of conceptual models based on their observations and perceptions of current reality (which can be among others captured in real world scenes, but also be only his memory). The abstraction process takes place in their minds and is by no means traceable. Thus, the influence of his observations on the definition of these concepts is not documented and important background information on which modelling decisions are based on are quickly forgotten and cannot be used for documentation and understanding of the models' rationale. In other words, important inherent relationships between the observations and concepts are not persistently recorded. Consequently, the group of people depicted in the upper right corner of Figure 1 (left) trying to review his creations can hardly assess the quality and correctness of the conceptual model.

Haumer Thesis Figure 1

Figure 1: Problem to be solved and solution idea: Providing pre-traceability to real world scenes for conceptual models.

The solution for this problem proposed by our approach is outlined in Figure 1 (right). We extend existing ethnographical and participatory design techniques using rich media (step (1) of Figure 1 (right)) with a goal-centred model elicitation and validation technique providing tool-supported traceability (step (2) of Figure 1 (right)). More concretely, when the analyst of Figure 1 uses real world scenes for the abstraction process (i.e. either for the election or validation of concepts), we provide him with (a) with guidance for the abstraction of concepts such as goals and (b) with support for a persistent interrelation of the concepts with the parts of the real world scenes that have influenced its definition.

The established interrelations result in a special form of pre-traceability which extends existing traceability approaches in the way that it provides i) traceability back to concrete instance examples from the real world instead of just tracing between different representations of abstractions and ii) traceability in a fine-grained way allowing interrelations of arbitrary parts of conceptual models with arbitrary parts of real world scenes and not just interrelationships on a document level. This enables fast and selective access to the relevant parts of the real world scenes, documenting how model components have been elicited or validated and avoids time-consuming viewing of irrelevant information.

For the interrelations we provide method and tool support to use them in a reference base kind of manner to support explanation, review, and negotiation of the conceptual models. The interrelated concepts are also used as a structuring mechanism for the (normally unstructured) real world scenes allowing, for instance, to compare different observations (e.g., two different usage situations of the same activity in different locations) in respect to the concepts they are related to.

Support for creation, maintenance, and methodical application of the interrelations requires appropriate tool support. This has been established in the PRIME-CREWS environment, a framework for process integrated modelling environments. PRIME offers generic and flexible mechanisms for supporting method adaptability as a basis for incorporating method guidance for establishing and applying the interrelations. At the moment the PRIME-CREWS environment offers method support for a multimedia editor allowing to manage and edit real world scenes, a conceptual goal editor allowing to edit and manage textual goal and requirements hierarchies, as well as a message sequence charts editor allowing to edit behavioural interaction models.

Bibliography

[Haumer et al., 1998a]
P. Haumer, K. Pohl, and K. Weidenhaupt. "Requirements Elicitation and Validation with Real World Scenes", in IEEE Transactions on Software Engineering, Vol. 24, No. 12, Special Issue on Scenario Management, 1998, pp. 1036-1054.

[Haumer et al., 1998b]
P. Haumer, P. Heymans and K. Pohl, "An Integration of Scenario-Based Requirements Elicitation and Validation Techniques", CREWS Report Series, No. 98-28, available at http://sunsite.informatik.rwth-aachen.de/CREWS/, 1998.

[Haumer et al., 1999a]
P. Haumer, K. Pohl, K. Weidenhaupt, M. Jarke, "Improving Reviews by Extended Traceability", Scenario-Based System Development in the Emerging Technologies Track of the 32nd Hawaii International Conference on System Sciences (HICSS-32), Jan. 1999; a revised and extended version is to appear in Interacting with Computers, 1999.

[Haumer et al., 1999b]
P. Haumer, J. Rack, K. Pohl, "Interrelating Goal Models and Multimedia Scenes: An Empirical Investigation", Proc. of 5th Int'l. Workshop on Requirements Eng.: Foundation for Software Quality (RESFQ99), Heidelberg, Germany, June 14-15, 1999, pp. 83-98.

[Haumer et al., 1999c]
P. Haumer, P. Heymans, M. Jarke, and K. Pohl, "Bridging the Gap Between Past and Future in RE: A Scenario-Based Approach", in Proc. of 4th IEEE International Symposium on Requirements Engineering (RE'99), University of Limerick, Ireland, 7-11 June 1999.

[Jarke et al., 1997]
M. Jarke, K. Pohl, P. Haumer, K. Weidenhaupt, E. Dubois, P. Heymans, C. Rolland, C. Ben Achour, C. Cauvet, J. Ralyté, A. Sut-cliffe, N. Maiden, S. Minocha, "Scenario Use in European Software Organizations - Results From Site Visits and Questionnaires", CREWS Report Series, No. CREWS-97-10, available at http://sunsite.informatik.rwth-aachen.de/CREWS/.

[Pohl et al., 1999]
K. Pohl, K. Weidenhaupt, R. Dömges, P. Haumer, M. Jarke, R. Klamma, "Process-Integrated (Modelling) Environments (PRIME): Foundation and Implementation Framework", to appear in ACM Transactions on Software Engineering and Methodology, 1999.

[Weidenhaupt et al., 1998]
K. Weidenhaupt, K. Pohl, M. Jarke, and P. Haumer. "Scenario Usage in System Development: A Report on Current Practice", IEEE Software, Mar., 1998, pp. 34-45.

Document Actions