Towards a component -based approach to software engineering for knowledge -intensive systems: Applying the task -structure approach to develop a knowledge-based system for the inherent risk assessment audit task

The knowledge modeling approach illustrated in the current research emphasizes identifying and understanding the various problem-solving roles of knowledge in the application task, during a conceptualization stage rather than beginning the system development at the representational level. Knowledge level modeling provides an additional level of system specification (the knowledge model) to identify the various roles that knowledge plays in problem solving. In the current study, knowledge level modeling is operationalized by adopting the Task-Structure modeling approach. This approach emphasizes (a) understanding the task properties and the organization structures of the required knowledge and (b) modeling of their relationships among knowledge elements into a knowledge model.;The current research examines the utility and benefits of using the Task Structure (TS) approach to design a knowledge intensive system for audit tasks, and specifically focuses on applying the TS approach to modeling the inherent risk assessment task in an auditing setting. This task was selected as a single aspect of a more general application of the approach to other audit tasks. This project examines the problems of shallow knowledge structure and the related validation issues experienced in the prior knowledge intensive systems projects by focusing on answering three questions: (1) How can knowledge be atomized and organized according to the roles it plays in problem solving? (2) Does the task-structure approach provide a parsimonious set of constructs for the above analysis? (3) Does the explication of the different knowledge primitives facilitate the validation of the knowledge intensive system developed?;In order to allow validation to be performed before a premature commitment to the production of a functional prototype, the knowledge model for the system under design should be delineated through a multi-stage evolutionary process. Thus, a major objective of this current research is to examine whether the contents of the knowledge model specified at the different stages of the evolutionary process facilitate a stage-wise validation. Towards this objective, experienced audit managers from one of the Big-Five audit firms were used to validate the system design based on documentation of the task structure analysis. The validated system design was then implemented and verified by the same audit managers.