| Abstract: | In recent years, a number of difficulties in designing interactions between military personnel and their command and control systems have been identified. These difficulties have been attributed to a lack of carry forward between procurement projects. This paper suggests that lack of carry forward is an integral part of current 'case by case' approaches to HCI. Consequently, a fundamentally different approach to HCI is required. The approach suggested here is a class approach. A class approach to HCI makes classinstance relationships between knowledge representations explicit by organising knowledge representations into class hierarchies. Given such hierarchies, procurement projects may consider the relevance of existing knowledge by attempting to locate the problem at hand within the hierarchy. Thus, a class approach to HCI may encourage carry forward by providing: (a) the opportunity to develop multiple instances of classes of interaction by specialising and instantiating class knowledge representations for the instances at hand; (b) the opportunity to apply research knowledge at different levels of development-to the development of the class and the instance (not just the case); and (c) an additional means of reasoning about the completeness/selectivity of instance knowledge representationswith respect to relevant, super-ordinate representations. This paper presents an initial illustration of a class approach to HCI. It identifies some key characteristics of a class approach to HCI, and then presents research and development work which exhibits these characteristics. Such an illustration is required, because current understanding about the nature of HCI concerns, and the relationships between HCI knowledge, practices and problems is such that one may not assume that all desirable approaches to HCI are necessarily realisable. Successful initial illustration provides an additional, encouraging precedent for full development of the approach. |
