All NewsML-G2 Items must have a @guid, an identifier that should be globally unique for all time and independent of location. The IPTC has registered a URN namespace for the purpose of creating GUIDs for NewsML-G2 Items using a specification based on RFC3085. The syntax for a @guid using this scheme is:

Use an internet domain name owned by your organisation as a ProviderId, for example:

A simple indicator of the version of the Item:

Version numbers need not be consecutive, but must START at 1, and a new version must be a higher number than the previous version. If version is missing, the value is assumed to be 1. See Hidden Values of NewsML-G2

A string denoting the IPTC Standard, in this case "NewsML-G2".

A string denoting the major and minor version of the Standard being used:

There are two levels of Conformance to the NewsML-G2 standard. The "Core" conformance level (CCL) represents a minimal sub-set of NewsML-G2 that can usefully get work done, and is the default if @conformance is omitted. In practice, many implementers use the properties of "Power" conformance (PCL). If an implementer stays within the CCL, the @conformance of the NewsML-G2 Item can be assumed and omitted, but must be declared if using PCL:

Setting the default language of XML elements with text content in NewsML-G2 is UK English:

Sets the default namespace for elements:

Putting this together, the required root element attributes are:

As previously mentioned, the <itemClass> property describes the type of content conveyed by the Item. It is mandatory to use one of the IPTC News Item Nature NewsCodes (recommended Scheme Alias: "ninat") for the Item Class of News Items and Package Items, expressed as a QCode:

Other possible values from this scheme include (not limited to) "ninat:picture", "ninat:video" and "ninat:audio".

Can be represented by a QCode, or a URI. If the value of this property is NOT taken from a controlled vocabulary, the @qcode or @uri will be omitted and the child <name> element used to give a human-readable value for the property. The IPTC recommends using a QCode with the Provider NewsCodes, a controlled vocabulary of providers registered with the IPTC with recommended Scheme Alias of "nprov":

This contains the date, time and time zone (or UTC) that this version of the NewsML-G2 Item was created. The value must be expressed as XML Schema datetime:
YYYY-MM-DDThh:mm:ss±hh:mm

The -05:00 denotes U.S. Eastern Standard Time offset from UTC

Every NewsML-G2 Item must have a publication status. The value defaults to "usable", which permits the <pubStatus> property to be omitted, however it is recommended that a value is explicitly included:

Publication status is highly likely to be used by most news agencies, because the ability to explicitly signal the status of news is essential. The use of the IPTC Publishing Status NewsCodes is mandatory. Its recommended alias is "stat". Other values permitted by the scheme are:

When an Item is cancelled, it must never be used, but an Item that has been withheld may subsequently have its status updated to "usable".

The <firstCreated> element indicates when the first version of the Item (not the content) was created.

Business-to-business news organisations often use an embargo to release information in advance, on the strict understanding that it may not be released into the public domain until after the embargo time has expired, or until some other form of permission has been given.

Embargoed is NOT the same as the Publishing Status; embargoed content should have a publishing status of "usable".

It is not required to give any further information about the embargo conditions, but some providers may provide a natural language <edNote>, see below.

The <service> element allows the provider to declare which of its services delivered this package, using a Controlled Vocabulary:

The <edNote> element contains a note that is intended to be read by internal staff at the receiving organisation, but not published to the end-user; in this example it conveys some optional information about the release condition imposed by the <embargoed> element:

Additional processing instructions can be given using <signal> and its @qcode. This example uses the IPTC Signal NewsCodes (recommended Scheme Alias "sig") that advises the end-user that this Item updates a previous versions of the Item:

The other value in the Signal scheme is "correction". There is further NewsML-G2 functionality for expressing fine-grained information about the reason and impact of updates and also for applying signals to different parts of an Item’s content.

The <link> element has two basic purposes:

This example provides a "see also" link to a resource on the Web that end-users can view to get further information about the event. @rel is used to denote the reason that the link is provided. In this example, the QCode uses the recommended IPTC Item Relation NewsCodes with a recommended Scheme Alias of "irel" and the code value is "seeAlso":

The <contentCreated> timestamp corresponds to a "Created on" field of the story. It is expressed in NewsML-G2 as Truncated DateTime data type, meaning that the date-time elements may optionally be stripped, starting from the right. If required, the <contentModified> property may also be used to contain the "Last Edit" timestamp. This must be later than the Created timestamp.

The place that the content was created uses the <located> element. Note that this is not necessarily the place of the event or subject. For example, for a UK story written in the London office, <located> would be "London"; a picture of Mount Fuji taken from downtown Tokyo would have a <located> value of "Tokyo".

The semantics of <located> are similar to the natural-language location carried in the dateline that often prefaces news (such as "BERLIN, October 24") but can be conveyed more precisely, and in terms that may be more readily processed by software, using a @qcode or @uri:

The optional @type uses a controlled vocabulary to indicate the nature of the location being expressed; in the example this indicates that <located> refers to a city.

Both Located and Subject (below) contain child elements that express a specific relationship between entities or concepts. For example, the content originated in the city of Berlin and the <located> element shows that the city of Berlin has a "broader" relationship – that is a child-to-parent relationship – to Berlin the state, and to Germany the country:

The writer, photographer or other party (person or organisation) who created the content is expressed using the <creator> element:

The <infoSource> element, together with its optional @role, enables finely-grained identification of the various parties who provided information used to create and develop an item of news. If absent, the default value of @role is the originator of the information used to create or enhance the content.

The subject matter of content uses the <subject> element. When the value of the Subject is taken from a Controlled Vocabulary, this is identified using either a @qcode or @uri:

For concepts not taken from a CV, the identifier is omitted and the name of the concept is given in the child <name> element, for example:

The optional @type uses the IPTC "nature of the concept" NewsCodes (recommended scheme alias "cpnat") to indicate the type of concept being expressed, for example, an abstract concept, that is a concept that does not represent a real-world entity, but something like an idea, or news category.

The above example uses a concept from the IPTC Media Topic NewsCodes. Also note the use of the W3C XML attribute xml:lang that expresses the language used for the element’s value. It is also possible to add relationships to related concepts, as shown above in <located>. For example:

The IPTC highly recommends that providers use the Media Topic NewsCodes unless there is an over-riding requirement to use proprietary codes. This promotes inter-operability and standardisation in news exchange. It is also recommended that if a <name> is used in conjunction with a QCode, then the value of the <name> agrees with the value in the Scheme for the language specified in xml:lang. IPTC NewsCodes are provided in UK English (en-GB) and translations in French and German have been provided by IPTC members.

The "medtop" code prefix is the recommended Scheme Alias for the Codes, which is resolved via the IPTC Catalog (see Catalog wrapper <catalogRef>, above) to the Scheme URI http://cv.iptc.org/newscodes/mediatopic/ This is guaranteed to be globally unique because it is part of the Internet Domain controlled by the IPTC. Appending the code "04000000" to the Scheme URI forms the Concept URI http://cv.iptc.org/newscodes/mediatopic/04000000 that cannot be confused with a concept with the same code 04000000 from another source.

When using @type to indicate the nature of the concept, the possible values from the IPTC Concept Nature NewsCodes are:

The <genre> element indicates the style of the content, in this example "interview" as a property that is distinct from <subject> that is used to indicate the subject matter of content. In the example, an IPTC Genre NewsCodes value of "Interview" is used:

Some news services implemented in NewsML-G2 retain the <slugline> property as a human-readable index for legacy reasons; therefore receivers may sometimes see this property. However, it has never been a completely reliable identifier, and it is recommended that more purposeful identifiers that are also machine-readable are implemented in its place.

Even if the Headline is carried inline in text content, it is useful also to place it explicitly in metadata so that it can more easily be identified and extracted by the end-user: