Context management in modeling information systems (IS)

Abstract

Knowledge engineering methods are shown to have an important role in addressing the challenge of building trusted contexts; namely, in providing solutions to the problems of complexity, conformity, changeability and invisibility. Current information systems (IS) design practices, however, are seen to solve only those problems pertaining to the global enterprise system, while leaving the validation issues related to user/local models unresolved. Context management mechanisms associated with the user/local models are shown to provide a basis for dynamic validation of user requirements. The proposed context management process has been developed to allow context representation and dynamic knowledge structuring at the epistemological level.

Introduction

The complexity of modern information systems (IS), particularly those designed for electronic commerce or enterprise-wide solutions, such as supply chain management, is based on the high levels of commercial integrity and performance now required to deliver value to customers in a globally competitive environment. The need to model the performance of these mission-critical systems and to establish that the requirements/contribution of all stakeholders have been accurately captured, validated and integrated in the design of the requisite system is, therefore, critical if the IS conceptual foundations and modeling methodologies are to be trusted.

Trusted systems have traditionally referred to highly secure, complex systems, typically in defence and the intelligence communities, and more recently, in electronic payment systems. Formal methods have often been used to specify such systems requirements, and real-world modeling is regarded as essential in validating real-time control and command environments of specified operational performance. More recently, formal methods have been used to model the requirements of complex commercial systems, such as corporate EDI gateways [21], but the high level of trust required of such systems can only be satisfied with effective mapping processes between the requirements specification and the IS design. With the disappearance of financial paper, such as share certificates, and more generally, the move to digital certification technologies for authentication and validation, and in evidence based medicine for example, there is an overarching requirement for trusted systems.

In 1987, Brooks [4] described four essential difficulties that are inherent in software products, namely, complexity, conformity, changeability and invisibility. As software products are, in fact, at the core of any computerized IS, these difficulties can also be identified as pertinent to enterprise-wide IS solutions. It is, therefore, necessary to extend the original descriptions to encompass IS systems and to ask what other properties are required for trusted systems?

Complexity: IS are inherently complex in terms of a number of elements and their states and interactions that are needed to achieve the system functions.

Conformity: IS must conform to the business and technical environments they operate in, otherwise they fail to provide the functionality that is expected of them.

Changeability: IS are embedded in a cultural matrix of business processes, users, laws and technology solutions that change continually, forcing changes upon the information systems.

Invisibility: Software is invisible and unable to be visualized. Therefore, all IS requirements that are not explicitly represented become hidden in the programs that implement them.These difficulties are not mutually exclusive. Invisibility makes it difficult to maintain conformity, as the original requirements are not apparent when validating the requirements or implementing necessary changes. Changeability is affected by complexity since it is more difficult to plan and design changes in a complex system. And of course, conformity must be continuously maintained in the ever-changing business and technical environments.

It is also important to understand that the four issues are equally relevant, both when developing an initial IS solution, and also, when IS modifications are forced by changing requirements. The design and development methods employed to build the system decide how easy or difficult it is to make subsequent modifications. Subsequent changes should also be expected to affect, usually adversely, the system's maintainability!

In this paper, contemporary IS development practices, addressing the difficulties cited earlier, are reviewed. In particular, it is shown that knowledge acquisition and representation techniques are being combined with traditional development methods to create more efficient development environments. It is argued that development techniques and knowledge modeling methods currently in use, focus on efficiency rather than the requirement for trusted system, and therefore, are inadequate to address invisibility and conformity problems. The need for contextual reference and context management is identified to resolve the underlying conflict between local and global requirements in building trusted systems.