Development of a Theory-Based Ontology of Design-Induced Error

Abstract

The research in this thesis focuses on investigating the possible influence of design of artefacts and systems on human error, and developing means by which the factors contributing to human error related to design can be better understood. The thesis suggests that integration of current theories related to human error and design issues can help designers to pick up design issues from human error cases. The motivation of the research was initiated by the non existence of a unique human error theory relating design issues and human error. These theories have never been systematically reviewed for their characteristics and relationships. The diversity of theories makes it difficult to identify design issues from human error cases. The premise that a collective model can be constructed with a paradigm is the basis of the research. The expression “design-induced error” in this thesis is used to refer to human error influenced by design.

This thesis proposes a meta-theory of design-induced error to provide an integrated and collective view of related theories by comparing their key characteristics and adopting a paradigm. The thesis then develops an ontology of design-induced error based on the meta-theory. The developed ontology is a reasoning support tool in knowledge-based systems aiming to capture and to recognise design issues in human error cases. The models developed were examined through the analysis of accident reports. The Australian aviation accident report system (AAARS) is the main source for the experiment.

The ontology developed of design-induced error describes concepts and relations that relate to capturing design issues from human error cases. The ontology examines terms in accident documents such as terms indicating human error and the role of design. The software for developing a knowledge model (e.g. PCPACK) enables us to break down documents into concepts and relations in order to pick up relevant terms effectively from documents.

The models developed in this research provide analytical means for designers as well as accident analysts to identify relevant design issues from accident cases or documents.

Date of Award	1 Jul 2009
Original language	English
Awarding Institution	University of Bath
Supervisor	Christopher McMahon (Supervisor) & J Busby (Supervisor)