Choosing a problem — when is Artificial Intelligence appropriate for the retail industry?

Abstract

Abstract: Current methodologies for choosing problems suitable for implementation as IKBS's are not appropriate for the retail industry, due to singularities of the domain. In particular, they exclude from consideration a kind of system which uses Electronic Point of Sale (EPOS) data to structure unstructured managerial perceptions, but several examples of these systems have been successfully implemented. Methodologies may be more domain specific than they were thought to be, and may hold back implementations to an artificially imposed level of difficulty. Nonetheless, analysis of the level of structure found in a problem is useful, as it points out possible trouble spots, and Fuzzy/Complex analysis was felt to be the most suitable methodology for retail due to its manufacturing origins, clarity of purpose, and flexibility.