Best practises

This is a list of Best Practises for Multilingual LOD. The information here comes from:

• Preliminary notes used for the group discussions

• (Draft) report about patterns for MLOD

• Minutes and emails of the community group

The list is deliberately concise and will serve as input for the W3C DWBP working group as well as for the final best practises community group report.

Naming

There is a variety of possible patterns for defining URIs, from a multilingual perspective (see General Patterns for Multilingual Linked Open Data): descriptive URIs, opaque URIs, full IRIs, path only IRIs, etc.). The use of one or another will largely depend on the legacy data being converted into LD and the type of application. General advise on URIs definition (such as the ones contained in the Best Practices for Publishing Linked Data) still applies in a multilingual context. In particular:

BP-N1	Agents making use of URIs/IRIs should not attempt to infer language properties or assume linguistic data encoded in the URI itself.
BP-N2	If IRIs are used it is preferable that an ASCII domain is still used (path-only IRIs).

Dereferencing

In principle we see few arguments in favour of language content negotiation in the context of Linked Data (http://www.w3.org/2014/02/21-bpmlod-minutes.html)

BP-D1

how to formulate a BP from this?

Textual information

BP-T1	Use rdfs:label whenever possible, even if richer or alternate linguistic descriptions are also available for the same entity (e.g., lemon, SKOS, …)
BP-T2	Use language tags in labels and textual descriptions whenever possible.