Using Extended Attributes in Data Analysis Software - Controlled Vocabularies, Tools and DDI

Abstract

All of the major data analysis software packages now allow some form of user defined extended attributes on variables and most also allow these attributes for the datasets themselves. In each case these attributes can be seen as a pair of strings (attribute name, attribute value). They can also be seen as a subject, predicate, object triple (variable, “has” attribute name, attribute value). This paper explores potential uses of these attributes and suggests directions for developing best practice guidelines for their use.