10.5. Classification

Most sets of data objects can be organized into smaller subsets as they reflect some commonalities, and master data objects are no exception. Master data objects can often be classified within logical hierarchies or taxonomies; sometimes these are based on predefined classes, whereas interestingly, other data sets are self-organized. For example, products may be segmented into one set of categories when manufactured, another set when transferred from the factory to the warehouse, and yet another set when presented for sale. This means that different applications will have different views of master data and suggests that entity classification enhances the ability to integrate applications with a master data repository.

Forcing a classification scheme requires a significant amount of prior knowledge of the objects, how their attribution defines the segmentation, and even the categories for classification. Many business clients can clearly articulate the taxonomies and classification criteria for the data their applications use, whereas in other environments the data can be subjected to an assessment and analysis in order to construct the hierarchies and categories.