An expert advisor system for college management.

This thesis has explored the economic, political, legal and technological changes that have an impact on decision support requirements in many organisations. It has looked particularly at the Public Sector and the FE Sector and has established the need for an intelligent decision support system. Critical Success Factors have been identified that have influenced the design of a specific Expert Advisor System experimental prototype, the development of which has been central to the research. A range of system development methodologies have been reviewed, and justification has been provided for the selection of the CommonKADS methodology. Technologies and techniques to be used in the development of the Expert Advisor System have also been reviewed and justification has been provided for incorporating Case-Based Reasoning and Data Warehousing. The analysis, design and development of the system has been strongly influenced by the Critical Success Factors that were identified to ensure the system met the decision support needs. The experimental prototype has been developed specifically to assist Senior Managers at an FE college with the decision making that is used to complete ISR Funding Returns. The system gives access to historic data, provides auditable data trails to substantiate decisions and facilitates what-if projections. Case-based knowledge discovery, data-based knowledge discovery, graph-based knowledge discovery and projection-based knowledge discovery have been achieved through the use of the prototype. An important part of the development process was the adaptation of cases and the adaptation of queries that extracted and aggregated data to provide system adaptation. The research has focused around addressing two research hypotheses. Evidence has been provided to show that the two research hypotheses have been addressed. This demonstrates that (hypothesis 1) CommonKADS Models are well suited to providing a template for the design and documentation of Decision Support Systems that need to operate in rapidly changing domains. Justification has also been given to show that (hypothesis 2) CBR principles can be used together with other knowledge discovery techniques to provide useful adaptive systems. The research concludes by looking at how new technologies could be incorporated in later versions of the Expert Advisor System.