Search

Context-aware case-based decision support systems (CACBDSS) use the context of users as one of the features for similarity assessment to provide solutions to problems. The combination of a context-aware case-based reasoning (CBR) with general domain knowledge has been shown to improve similarity assessment, solving domain specific problems and problems of uncertain knowledge. Whilst these CBR approaches in context awareness address problems of incomplete data and domain specific problems, future problems that are situation-dependent cannot be anticipated due to lack of data by the CACBDSS to make predictions. Future problems can be predicted through situation awareness (SA), a psychological concept of knowing what is happening around you in order to know the future. The work conducted in this thesis explores the incorporation of SA to CACBDSS. It develops a framework to decouple the interface and underlying data model using an iterative research and design methodology. Two new approaches of using situation awareness to enhance CACBDSS are presented: (1) situation awareness as a problem identification component of CACBDSS (2) situation awareness for both problem identification and solving in CACBDSS. The first approach comprises of two distinct parts; SA, and CBR parts. The SA part understands the problem by using rules to interpret cues from the environment and users. The CBR part uses the knowledge from the SA part to provide solutions. The second approach is a fusion of the two technologies into a single case-based situation awareness (CBSA) model for situation awareness based on experience rather than rule, and problem solving predictions. The CBSA system perceives the users’ context and the environment and uses them to understand the current situation by retrieving similar past situations. The futures of new situations are predicted through knowledge of the history of similar past situations. Implementation of the two approaches in flow assurance control domain to predict the formation of hydrate shows improvements in both similarity assessment and problem solving predictions compared to CACBDSS without SA. Specifically, the second approach provides an improved decision support in scenarios where there are experienced situations. In the absence of experienced situations, the second approach offers more reliable solutions because of its rule-based capability. The adaptation of the user interface of the approaches to the current situation and the presentation of a reusable sequence of tasks in the situation reduces memory loads on operators. The integrated research-design methodology used in realising these approaches links theory and practice, thinking and doing, achieving practical as well as research objectives. The action research with practitioners provided the understanding of the domain activities, the social settings, resources, and goals of users. The user-centered design process ensures an understanding of the users. The agile development model ensures an iterative work, enables faster development of a functional prototype, which are more easily communicated and tested, thus giving better input for the next iteration.