IMS VDEX - Technical Details

Technical Details

The VDEX Information Model is represented in the diagram. A VDEX file describing a vocabulary comprises a number of information elements, most of which are relatively simple, such as a string representation of the default (human) language or a URI identifying the value domain (or vocabulary). Some of the elements are ‘containers’ – such as a term – that contain additional elements.

Elements may be required or optional, and in some cases, repeatable. Within a term, for example, a description and caption may be defined. Multiple language definitions can be used inside a description, by using a langstring element, where the description is paired with the language to be used. Additional elements within a term include media descriptors, which are one or more media files to supplement a term’s description; and metadata, which is used to describe the vocabulary further.

The relationship container defines a relationship between terms by identifying the two terms and the specifying type or relationship, such as a term being broader or narrower than another. The term used to specify the type of relationship may conform to the ISO standards for thesauri.

Vocabulary identifiers are unique, persistent URIs, whereas term or relationship identifiers are locally unique strings. VDEX also allows for a default language and vocabulary name to be given, and for whether the ordering of terms within the vocabulary is significant (order significance) to be specified.

A profile type is specified to describe the type of vocabulary being expressed; different features of the VDEX model are permitted depending on the profile type, providing a common grammar for several classes of vocabulary. For example, it is possible, in some profile types, for terms to be contained within one another and be nested, which is suited to the expression of hierarchical vocabularies. Five profile types exist: lax, thesaurus, heirarchicalTokenTerms, ‘glossaryOrDictionary’ and flatTokenTerms. The lax profile is the least restrictive and offers the full VDEX model, whereas the flatTokenTerms profile is the most restrictive and lightweight.

VDEX also offers some scope for complex vocabularies, assuming the existence of a well-defined application profile (for exchange interoperability). Some examples are:

• Faceted schemes – faceted vocabularies are possible with the definition of appropriate relationships
• Multi-lingual thesauri – metadata could be used within a relationship to achieve multilingual thesauri
• Polyhierarchical taxonomies – can be expressed using the source/target value pairs in the relationship.

Identifiers in VDEX data should be persistent, unique, resolvable, transportable and URI-compliant. Specifically, vocabulary identifiers should be unique URIs, whereas term and relationship identifiers should be locally unique strings.