Críticas Cognitivas a Heurísticas Orientadas a Modelo

Abstract

The requirements, as the foundation stone of the software system development, must be of the best possible quality, covering the real needs of the clients and the limitations of the project. Reaching an acceptable level of quality is not easy and sometimes it is not usually met. In previous research projects, these quality levels have been thoroughly evaluated in natural language written models, mainly in glossaries with definitions of the terms of the language used in the context of application, observing in these cases a low quality and low repetitiveness thereof. This feature is virtually independent of the model and of the sources of information used. Despite repeated efforts consisting of the incorporation of new heuristics, the improvements achieved were unsatisfactory. A retrospective review of these works allowed to detect that it is undoubtedly the cognitive ability of the own requirements engineer the origin of these defects. Notably, the heuristics used provide rudimentary guides on how to prepare the elicited information. This lack is the basis of the exploratory-descriptive research project that has been initiated and presented here.