The process of research investigations in artificial intelligence-a unified view

A number of research communities recognize artificial intelligence (AI) as a valid reference discipline. However, several papers have criticized AI's research methodologies. This paper attempts to clarify and improve the methods used in AI. Definitions are proposed for terms such as AI theory, principles, hypotheses, and observations. Next, a unified view of AI research methodology is proposed. This methodology contains a long term dimension based upon the scientific method and an individual project dimension. The individual project dimension identifies four strategies: hypothetical/deductive, hermeneutical/inductive, case-based, and historical analysis. The strategies differ according to how prototyping is used in an experiment.<>