Abstract

Metadata specifications, especially those found in standards such as DDI Lifecycle, GSIM, and SDMX, are large. They cover a range of descriptive needs. For those who build the standards, each section in them is understood well. But for users of the standards, interpreting and using them consistently from one application to the next is sometimes hard to achieve.

An example arises from the use of Study in DDI Lifecycle. Study is used to describe a single collection of data, all the collections in an ongoing survey, and sections of subject matter a survey covers. Humans can easily make sense of these differences; difficulties arise when metadata are compared automatically. Computers can’t make these distinctions unless they are told in advance.

These inconsistent decisions cause interoperability problems. Interoperability, however, is a complex topic and is often used without a full understanding of what it means. In general, and loosely speaking, it refers to the capability of using a system or some data without having to consult a third party.

Adding to the complexity, there are several kinds of interoperability: system, structural, syntactic, and semantic. System interoperability is self-explanatory. Structural interoperability refers to the ways data are organized. Syntactic interoperability addresses the representations of data. Semantic interoperability is about the meaning of data. Problems with semantic and structural interoperability are the issues the inconsistent use of specifications cause.

This paper describes the issues in more detail.