18. Designing a NewsML-G2 feed from scratch

When designing a NewsML-G2 feed, some fundamental questions need to be asked, such as:

This chapter gives general guidance on the issues an implementer of a feed might encounter, and where necessary points to more detailed information on specific topics described elsewhere in the Guidelines.

As a general principle, the IPTC recommends that providers strive for inter-operability, so where possible, stick to using the recommended "mainstream" properties as identified by the Property Usage sheet below. The sheet lists the NewsML-G2 properties which are most widely used by currently existing NewsML-G2 feeds. As there may be more than one way to implement a feature; in such cases the IPTC recommends choosing the most widely-used property as indicated by the sheet.

Also consider using existing IPTC NewsCodes Schemes for property values where appropriate, in preference to creating proprietary schemes that accomplish a similar purpose. A table showing IPTC Controlled Vocabulary Usage is shown below.

The design of a NewsML-G2 feed will depend on the content type being conveyed, since each type will have specific metadata and processing needs.

For all content types, note the basic NewsML-G2 rule: a single logical item of content per News Item, which can be represented by more than one technical representation. For multiple pieces of content in a named relationship, use the Package Item, and for sending one or more NewsML-G2 Items in a transmission workflow, the News Message may be used.

The Chapter on NewsML-G2 Basics has a description of the mandatory and commonly-used properties.

Choose a scheme for generating the @guid of NewsML-G2 Items. The IPTC has registered a namespace urn "newsml" and a specification for this purpose (see Item Identifier) but other schemes such as TAG URI are available.

Decide how Item version will be generated: it must start at "1" and ideally should increment by "1" each time a new version is published, although a new version may have any number that is greater than the previous version number. Providers may also use <link> to inform receivers of the previous version and provide a location or id for it (see Using Links for more details on this)

Since version 2.24, NewsML-G2 supports only one schema (the "All-Power" schema), therefore the conformance attribute must be present with a value of "power".

When designing and testing the feed, validation against the IPTC scheme files will be essential. The IPTC recommends that validation should not be required in production: it can be resource intensive and introduces a potential point of failure. If a provider chooses to validate production documents, they must NOT validate against the schema files hosted by the IPTC.

Even after following the advice about using NewsCodes (above) it is almost certain that some provider specific schemes will be needed. See the section on Controlled Vocabularies and QCodes for detailed information on how to create and maintain CVs and Catalogs.

See Chapter on Rights Metadata to go beyond the basics of expressing rights information.

The following property usage table was compiled from a survey of IPTC members.

The table below compares the usage of IPTC CVs (where available) and proprietary CVs for commonly-used and mandatory properties, compiled from a survey of IPTC members.