Preview. Built from the published NewsML-G2 documents by tools/prosedoc. This layout is a harness for reviewing structure — it is not the published site.
20. Mapping Embedded Photo Metadata to NewsML-G2
20.1. Introduction
Embedded metadata in JPEG and other file formats have been a de facto standard since the 1990s. In fact, the metadata schema adopted by Adobe Systems Inc for the original "File Info" dialog box in Photoshop is based on the IPTC Information Interchange Model (IIM). This standard for exchanging all types of news assets and metadata pre-dates XML-based standards such as NewsML-G2. The IIM-based embedded properties in images became known as the "IPTC Fields" or "IPTC Header" and they were widely adopted in professional workflows.
In about 2001, in order to overcome some technical limitations imposed by this legacy model, Adobe introduced the Extensible Metadata Platform, or XMP, to its suite of applications that includes Photoshop. Adobe also worked with the IPTC to migrate the properties of the legacy "IPTC Header" to XMP. Most of the original IIM-based metadata properties are contained in the IPTC Core Schema for XMP. Further IPTC properties that are derived from NewsML-G2 properties are available in the IPTC Extension Schema for XMP.
| Although developed by Adobe, XMP is an open technology like its IIM-based predecessor, and has been adopted by other software vendors and manufacturers. Adobe and others, including Microsoft, Apple and Canon, have formed the Metadata Working Group, a consortium that promotes preservation and inter-operability of digital image metadata. The MWG web site is www.metadataworkinggroup.org. |
However, when IIM-based properties were added to software packages such as Photoshop, the original semantic definitions of IIM were left behind, and in the absence of formal definitions, the use of these properties among vendors and agencies began to diverge as semantics of the labels were interpreted differently. The IPTC addressed this in 2005 by bringing interested parties together to agree the IPTC Photo Metadata Standard, based on recognised current industry practice. The latest version (as of this edition of the Guidelines) was approved in 2016. (For more information on the Standard visit the IPTC website www.iptc.org and navigate to Photo Metadata.)
A mapping of IIM legacy properties to NewsML-G2 is given below in IIM to NewsML-G2 Field Mapping Reference, but this can only be regarded as "the basics". A full mapping of the IPTC Photo Metadata Standard backwards to IIM (where feasible) and forwards to NewsML-G2 may be found at IPTC Photo Metadata Mapping on the IPTC website. Note that because the photo metadata world has moved on since the original IIM specification, the semantic interpretation of IIM in the mapping guide follows the Photo Metadata Standard, which is in some cases slightly different to the IIM 4.2 Specification.
20.2. IPTC Metadata for XMP
To use the IPTC Photo Metadata in Adobe CS3 and CS4 a set of custom panels for Bridge must be downloaded from the IPTC web site and installed on your Windows or Mac OS X computer. Adobe CS5 and higher provides built-in File Info panels covering both IPTC Core and Extension. Be aware that these custom panels only work up to version CS 5.5 of Bridge. To download visit www.iptc.org and navigate to Photo Metadata.
The IPTC web site also lists image manipulation software packages that synchronize the legacy IIM-based header fields and XMP properties. The IPTC Custom Panels User Guide tabulates the field labels used by IIM, IPTC Core, and the equivalent fields of software packages such as Photoshop, iViewMedia, Thumbs, and Irfanview.
20.3. Synchronizing XMP and legacy metadata
When opening an image that contains the legacy IIM-based IPTC metadata, using an application that supports XMP, users may need to be aware of synchronization issues. Digital images that pre-date XMP contain the legacy IIM-based metadata in a special area of the JPEG, TIFF or PSD file called the Image Resource Block (IRB). XMP-aware applications store metadata in the file differently, in the XMP Packet.
Figure: The IPTC Core metadata panel for an image opened in Adobe Photoshop.
Current versions of Adobe’s software such as Photoshop write the metadata to the IIM-header wrapped by the IRB, and to the XMP packet, when saving a file. Therefore, an image which contained only legacy metadata when opened could have metadata written to a new XMP packet (where an IIMXMP mapping exists) when saved.
If a picture with XMP metadata is modified by a non-compliant application (i.e. the IRB only is changed), this should be detected by the next XMP-compliant processor that opens the picture, using a checksum. The Metadata Working Group guidance document has a detailed chapter on these synchronization issues.
There is no guarantee that the synchronization of legacy and XMP metadata will be supported in future versions of Adobe software, or applications from other vendors.
20.4. Picture services using IIM/XMP
Many news and picture agencies deliver images to their customers using IIM. Customers receive these services in three variants:
-
As binary files such as JPEG that support embedded IIM metadata in the file header as "IPTC Fields" and/or XMP.
-
As binary files such as JPEG with additional embedded IIM fields beyond the set adopted by Adobe that are used by proprietary image-management software applications. These fields are not read by off-the-shelf software packages such as Photoshop. This format is used by the photo departments of many news agencies.
-
As binary IIM files which consist of the IIM envelope and associated object (picture) data. The conveyed picture may also contain embedded IPTC-IIM fields, but this metadata may not be synchronized with the IIM envelope in all instances. Providers’ synchronization practices vary.
20.5. Rationale for moving to NewsML-G2
It is probable that providers will continue to embed metadata in image and graphics files, but both providers and customers may derive benefits from exchanging this content using NewsML-G2:
-
metadata carried in a NewsML-G2 document is accessible without the need to retrieve and open the associated image file;
-
in some business applications, editors have to modify picture metadata but have no access or permission to modify the metadata in the image file, and the provider needs therefore to instruct receivers to use the NewsML-G2 metadata;
-
partners in an information exchange can standardise on a common method for managing all kinds of news objects.
-
Some news providers’ picture services are already being migrated to NewsML-G2. Some are already using NewsML-G2 as their internal standard for storing and exchanging all types of news objects; it makes sense to migrate to a common standard for customer-facing applications too.
-
IIM is a legacy format and has some inherent issues, such as limits on field lengths and difficulty with internationalization.
20.6. IIM resources
The IPTC web site has an IIM Home page. Visit www.iptc.org/standards and follow the link to IIM to view the Specification and other documents. These include a list of software packages that support IIM metadata fields, and a link to a document that maps IIM fields IPTC Core (XMP) Software Package labels, maintained by David Riecks at www.controlledvocabulary.com.
More about the use of IIM for photos can be found in the "Photo Metadata" section of the IPTC website.
20.7. Approach
Although IIM can handle any type of media object, its use for media types other than pictures is now rare, so this section will focus on the issue of migrating IIM picture metadata, chiefly those found in Adobe’s Photoshop "IPTC Header", to NewsML-G2 properties. It is NOT intended to describe a mapping of the complete IIM envelope or IPTC Photo Metadata Standard to NewsML-G2. (Available at IPTC Photo Metadata Mapping)
20.8. IIM to NewsML-G2 Field Mapping Reference
IIM is organised into Records and DataSets. The DataSets that are embedded in image files are in the Application Record (Record Two). Each IIM DataSet is labelled according to the parent Record and its position within the Record (these numbers are not necessarily contiguous). The IIM Specification contains details of the record structure.
The table below shows IIM DataSets that have been (with noted exceptions) adopted for the IPTC Photo Metadata. Each DataSet is shown with its IIM name, equivalent IPTC Photo Metadata Name (if available), and the corresponding NewsML-G2 Property.
| DataSet | IIM Name | IPTC Photo Metadata Name | NewsML-G2 Property |
|---|---|---|---|
2:05 |
Object Name |
Title |
itemMeta/title |
2:10 |
Urgency |
(Deprecated) |
contentMeta/urgency |
2:15 |
Category |
(Deprecated) |
contentMeta/subject |
2:20 |
Supplemental Category |
(Deprecated) |
contentMeta/subject |
2:25 |
Keywords |
Keywords |
contentMeta/keyword |
2:40 |
Special Instruction |
Instructions |
itemMeta/edNote Alternative: rightsInfo/usageTerms |
2:55 |
Date Created |
Date Created |
contentMeta/contentCreated |
3:20 |
By-Line |
Creator |
contentMeta/creator/name |
3:25 |
By-line Title |
Creator’s Jobtitle |
contentMeta/creator/related |
3:30 |
City |
City (legacy) |
if interpreted as city of the location the photo was taken contentMeta/located/POIDetails/address/locality/name OR if interpreted as city of the location shown contentMeta/subject -→ assert/POIDetails/address/locality/name |
3:32 |
Sub-location |
Sublocation (legacy) |
if interpreted as sublocation of the location the photo was taken contentMeta/located/POIDetails/address/line OR if interpreted as sublocation of the location shown contentMeta/subject -→ assert/POIDetails/address/line |
3:35 |
Province/State |
Province or State (legacy) |
if interpreted as province of the location the photo was taken contentMeta/located/POIDetails/address/area/name OR if interpreted as province of the location shown contentMeta/subject -→ assert/POIDetails/address/area/name |
3:40 |
Country/Primary Location Code |
Country Code (legacy) |
if interpreted as country of the location the photo was taken contentMeta/located/POIDetails/address/country/@uri OR if interpreted as country of the location shown contentMeta/subject -→ assert/POIDetails/address/country/@uri |
3:41 |
Country/Primary Location Name |
Country (legacy) |
if interpreted as country of the location the photo was taken contentMeta/located/POIDetails/address/country/name OR if interpreted as country of the location shown contentMeta/subject -→ assert/POIDetails/address/country/name |
3:43 |
Original Transmission Reference |
Job ID |
itemMeta/altId If the reference is a Job ID use itemMeta/ memberOf |
3:45 |
Headline |
Headline |
contentMeta/headline |
3:50 |
Credit |
Credit Line |
contentMeta/creditline |
3:55 |
Source |
Source |
contentMeta/infoSource |
3:56 |
Copyright Notice |
Copyright Notice |
rightsInfo/copyrightNotice |
4:00 |
Caption/Abstract |
Description |
contentMeta/description |
4:02 |
Writer/Editor |
Description Writer |
contentMeta/contributor (with @role to indicate the role of the contributor |
Note: deprecated properties are no longer supported in IPTC Photo Metadata |
20.9. IIM to NewsML-G2 Mapping Example
The screenshot below shows fields of the IPTC Panel in Adobe Photoshop’s File Info dialog, containing metadata from the IPTC Core metadata schema. Note the neighbouring tab labelled IPTC Extension that contains metadata fields of the IPTC Extension schema.
Figure: IPTC fields in Photoshop File Info panel (photo and metadata ©Thomson Reuters)
The table below maps the IIM DataSets from the example to NewsML-G2. In the succeeding paragraphs, the mapping discussed in more detail.
| DataSet | IIM Name | Value | Notes |
|---|---|---|---|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
20.9.1. Object Name / Title
The <title> element is a child of <itemMeta> and is intended to be a human-readable identifier. The <itemMeta> block also contains some mandatory NewsML-G2 elements, as shown:
<itemMeta>
<itemClass qcode="ninat:picture" />
<provider qcode="ex-prov:acmenews" />
<versionCreated>2025-10-19T02:20:00Z</versionCreated>
<pubStatus qcode="stat:usable" />
<fileName>2016-09-21T152911Z_1417736630_S1BEUCPFFZAB_USA-CRASH-MILITARY.xml</fileName>
<title>USA-CRASH/MILITARY</title>
<edNote />
</itemMeta>Note that <fileName> contains the provider’s recommended filename for this G2 Item.
Some providers may additionally map this DataSet to the NewsML-G2 <slugline> property in <contentMeta>:
<slugline>USA-CRASH/MILITARY</slugline>20.9.2. Urgency
The DataSet Urgency has been deprecated in the latest version of the IIM, and in the IPTC Photo Metadata Standard. It may still be used in some services, and therefore if present may be mapped to the corresponding NewsML-G2 property, which is a child of <contentMeta>:
<urgency>4</urgency>20.9.3. Keywords
The semantics of keyword are somewhat open: some providers use keywords to denote "key" words that can be used by text-based search engines; some use "keyword" to categorise the content using mnemonics, amongst other examples. The IPTC recommends that the IIM/Photo Metadata Keywords property is mapped to the repeatable <keyword> element in NewsML-G2 (one <keyword> element per keyword)
If this mapping cannot be assumed, the IPTC suggests the following rules when configuring a mapping of Keywords metadata:
-
Assess if any existing NewsML-G2 properties align to this use of keyword. Typical examples are
-
Genres ("Feature", "Obituary", "Portrait", etc.)
-
Media types ("Photo", "Video", "Podcast" etc.)
-
Products/services by which the content is distributed
-
-
If the keyword expresses the subject of the content it MAY go into the <subject> property with the keyword string placed in a <name> child element of the subject with a language tag if required.
-
If migrated to <subject>, providers should also consider:
-
Adding @type if the nature of the concept expressed by the keyword can be determined
-
Using a QCode if there is a corresponding concept in a controlled vocabulary
-
If none of the above conditions are met, then implementers should default to using <keyword> with a @role if possible to define the semantic of the keywords.
The contents of the Keywords field in the example shown are intended to be used as natural-language "key" words that could be used by a text-based search engine to index the content. Therefore, they will be mapped to <keyword>.
<keyword role="ex-krole:index">us</keyword>
<keyword role="ex-krole:index">military</keyword>
<keyword role="ex-krole:index">aviation</keyword>
<keyword role="ex-krole:index">crash</keyword>
<keyword role="ex-krole:index">fire</keyword>In IIM, the Keywords DataSet is repeatable, with each holding one keyword; therefore each keyword is mapped to separate <keyword> properties, even though they may appear as a comma-separated list in software application dialogs.
20.9.4. Special Instruction
The contents of this field could go into <edNote>, a child of <itemMeta>, which is placed after the <title> element (if present), if the nature of the instruction is a generic message to the receiver or its nature is unknown:
<itemMeta>
...
<edNote>NO ARCHIVAL OR WIRELESS USE</edNote>
</itemMeta>If appropriate and advised by the provider, an alternative mapping for the contents of this field MAY be <usageTerms>, parts of the <rightsInfo> block:
<rightsInfo>
<copyrightHolder qcode="ex-prov:TR">
<name>Thomson Reuters</name>
</copyrightHolder>
<copyrightNotice>Copyright 2025 Thomson Reuters. </copyrightNotice>
<usageTerms>NO ARCHIVAL OR WIRELESS USE</usageTerms>
</rightsInfo>20.9.5. Date Created
This maps to <contentCreated>, a child of <contentMeta>, since it refers to the content itself, for example:
<contentMeta>
<contentCreated>2008-12-08</contentCreated>
...
</contentMeta>When there is a Time Created value present in the IIM Record, this should be merged with Date Created, as the NewsML-G2 property accepts a Truncated DateTime value (i.e. the value may be truncated if parts of the Date-Time are not available).
| If present, the time part must be used in full, with time zone, and ONLY in the presence of the full date. |
<contentMeta>
<contentCreated>2008-12-08T17:30:00-08:00</contentCreated>
...
</contentMeta>20.9.6. By-line, Credit, Source
These three IIM DataSets are complementary, but each has a distinct application:
-
By-line is intended to identify the creator of the content
-
Credit identifies the provider of the content
-
Source holds the identity of a party having a role in the supply chain of the content.
By-line
The recommended mapping for By-line (IIM 2:80) is to the <creator> child element of <contentMeta>, rather than <by>. This is because <creator> is an administrative property that is intended to be machine-readable; the IPTC recommends that controlled vocabularies should be used if possible. The NewsML-G2 <by> property is a human-readable natural language property that is intended for display, but does not unambiguously identify the creator.
The example below shows this identification metadata in its administrative context. Expressed in this way using QCodes, the metadata can be used for administration and search. Using a CV, the photographer can be uniquely and unambiguously identified. The optional <name> is shown, and the <creator> property also allows the use of the child element <related> which in this case is used to express the photographer’s job title, again using a QCode, from IIM DataSet 2:85 (By-line Title)
<creator qcode="ex-pers:JS001">
<name>John Smith</name>
<related rel="ex-personrel:jobtitle"qcode="ex-stafftitles:photo">
<name>Staff Photographer</name>
</related>
</creator>Credit
Credit should be mapped to the <creditline> child property of <contentMeta>. There is a <provider> property of <itemMeta>, but the Credit does necessarily reflect the provider. Many picture providers use IIM Credit to display the name of the person, organisation, or both, who should be credited when the picture is used. In this context, <creditline> is appropriate because it is a natural-language label that is intended to be displayed.
<creditline>Acme/John Smith</creditline>Source
In the original IIM specification up to 4.1, Source refers to the initial holder of the copyright. In IIM 4.2, the semantics of Source were changed to be "…a person or party who has a role in the content supply chain. This could be an agency, a member of an agency, an individual or a combination. Source could be different from Creator and from the entities in the Copyright Notice.". This more "open" definition reflects current usage by the broad range of users and aligns with the Photo Metadata Standard.
In some distribution systems, the original owner of the copyright (as distinct from the current owner) is important, and some providers use the Source field for this information. This was the original intention of the IIM Specification which used Source for the Original Owner, and the Copyright Notice (DataSet 2:116) to hold the copyright statement which includes the Current Owner.
Using the current definition of Source, the recommended mapping to
NewsML-G2 is the <infoSource> element, a child of <contentMeta> (OR if
the source is the original copyright owner
<rightsInfo> (with @validto time delimiter) with a <copyrightHolder>
child element).
In this case, the original copyright holder is Thomson Reuters, and the Source of the content (an image in this example) is Acme News. The following uses a controlled value to indicate the @role of the party named as the Information Source:
<contentMeta>
...
<infoSource role="cpprole:imgSupplier">
<name>Acme News LLC</name>
</infoSource>
...For more on this topic, see Identifying Sources and Workflow Actors
20.9.7. City, Province/State and Country
As discussed in Pictures and Graphics, geographical metadata in images may have different contexts:
-
The location from which the content originates, i.e. where the camera was located. NewsML-G2 has a <located> property to express this.
-
The location shown in the picture – in NewsML-G2 this is a <subject> property of the picture.
Although for the majority of pictures, these are effectively the same spot, one can envisage situations where these two semantically distinct locations are not in the same place: a picture of Mount Fuji taken from downtown Tokyo is one example which is often quoted.
When mapping from IIM or IIM-based Photoshop fields, we assume that the intention of the provider is to express the location shown in the image, as this is the more customary use of these fields. (Be aware that specific providers may have their own convention, and receivers are advised to check).
The location shown in the example image may be expressed using a single <subject> property for the location, linked to an <assert> block:
<contentMeta>
...
<subject type="cpnat:poi" literal="int001" />
...
</contentMeta>
<assert literal="int001">
<POIDetails>
<address>
<line>University City</line>
<locality qcode="ex-city:int002" type="ex-cpnatexp:geoAreaCity">
<name>San Diego</name>
</locality>
<area qcode="ex-state:int003" type="ex-cpnatexp:geoAreaProvState">
<name>California</name>
</area>
<country qcode="iso3166-1a2:US" type="ex-cpnatexp:geoAreaCountry">
<name>United States</name>
</country>
</address>
</POIDetails>
</assert>Using <assert> in this way can be an advantage if a concept is used more than once in the NewsML-G2 Item. For example if both the Location Shown and the Location Created are the same place, all of the required concept details can be grouped in one <assert> and shared by both properties, for example:
<located type="cpnat:poi" literal="int001" />
<!-- Camera location -->
...
<subject type="cpnat:poi" literal="int001" />
<!-- Subject of picture -->
...
<assert literal="int001">
<POIDetails>
...20.9.8. Original Transmission Reference
This is defined in IIM as "a code representing the location of original transmission", but in common usage this DataSet has a broader use as an identifier for the purpose of improved workflow handling (IPTC Photo Metadata: Job ID). These uses include:
-
As an identifier for the picture (perhaps on some content management system)
-
As an identifier for a series of pictures, of which this one is part, e.g. a group of pictures of the same event.
The first use should be mapped to the NewsML-G2 property <altId>, a child of <contentMeta>. <altId> has two attributes: @type indicates the context of the identifier using a QCode, and @environment indicates the business environment in which the identifier can be used. This is expressed using one or more QCodes (QCode List).
<altId type="ex-idtype:systemRef" environment="ex-acmesys:mdn ex-acmesys:iim">SAD02</altId>If the DataSet represents a Job ID, the recommended mapping is to the NewsML-G2 <memberOf> property, a child of <itemMeta>:
<memberOf type="ex-ref:jobref">
<name>SAD02</name>
</memberOf>20.9.9. Headline
Maps to the <headline> child of <contentMeta>, a block type element:
<headline xml:lang="en-US">A firefighter walks past the remains of a military jet that crashed into homes in the University City neighborhood of San Diego
</headline>Note the use of @xml:lang to declare the language and variant "en-US".
20.9.10. Copyright Notice
These fields correspond to the <copyrightNotice> element, a child of the <rightsInfo> block.
<copyrightNotice>
Copyright 2025 Thomson Reuters
</copyrightNotice>20.9.11. Caption/Abstract
The contents of this field are placed in the <description> element, part of <contentMeta> with a @role attribute to denote that the description is a picture caption:
<description xml:lang="en-US" role="drol:caption">
A firefighter in a flameproof suit walks past the remains of a military jet that crashed
into homes in the University City neighborhood of San Diego, California December 8, 2008.
The military F-18 jet crashed on Monday into the California neighborhood near San Diego
after the pilot ejected, igniting at least one home, officials said.
</description>20.9.12. Writer/Editor
The caption writer is often a different person to the photographer, so aligns with the NewsML-G2 <contributor> property, a child of <contentMeta>. If possible, use a QCode value to unambiguously identify the contributing person and a @role to describe their role in the workflow. This property may be extended to include contact details and other information such as job title.
<contributor role="cpprole:descrWriter" qcode="ex-pers:CK" />LISTING 22: Embedded photo metadata fields mapped to NewsML-G2
The listing below combines the examples above into a complete listing. The following options have been used:
-
<usageTerms> used for Special Instructions, instead of <edNote>
-
<assert> used in conjunction with <subject> to express the location shown in the image.
-
<altId> used for the Original Transmission Reference (instead of <memberOf>)
All Scheme Aliases used in the listing below indicate IPTC NewsCodes vocabularies, except for the following alias values (note that they are prefixed "ex-" to aid readability): ex-acmesystem, ex-cat, ex-city, ex-cpnatexp, ex-idtype, ex-krole, ex-pers, ex-personrel, ex-prov, ex-stafftitles, ex-suppcat and ex-state.
<?xml version="1.0" encoding="UTF-8" standalone="yes"?>
<newsItem xmlns="http://iptc.org/std/nar/2006-10-01/"
xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
xsi:schemaLocation="http://iptc.org/std/nar/2006-10-01/
./NewsML-G2_2.35-spec-All-Power.xsd"
guid="tag:acmenews.com,2008:WORLD-NEWS:USA20081208098658"
version="18"
standard="NewsML-G2"
standardversion="2.35"
conformance="power">
<catalogRef
href="http://www.iptc.org/std/catalog/catalog.IPTC-G2-Standards_41.xml" />
<catalogRef href="http:/www.acmenews.com/customer/cv/catalog4customers-1.xml" />
<rightsInfo>
<copyrightHolder qcode="ex-prov:TR">
<name>Thomson Reuters</name>
</copyrightHolder>
<copyrightNotice>
Copyright 2025 Thomson Reuters
</copyrightNotice>
<usageTerms>NO ARCHIVAL OR WIRELESS USE</usageTerms>
</rightsInfo>
<itemMeta>
<itemClass qcode="ninat:picture" />
<provider qcode="ex-prov:acmenews" />
<versionCreated>2025-10-19T02:20:00Z</versionCreated>
<pubStatus qcode="stat:usable" />
<fileName>2016-09-21T152911Z_1417736630_S1BEUCPFFZAB_USA-CRASH-MILITARY.xml</fileName>
<title>USA-CRASH/MILITARY</title>
<edNote />
</itemMeta>
<contentMeta>
<urgency>4</urgency>
<contentCreated>2008-12-08T17:30:00-08:00</contentCreated>
<infoSource role="cpprole:imgSupplier">
<name>Acme News LLC</name>
</infoSource>
<creator qcode="ex-pers:JS001">
<name>John Smith</name>
<related rel="ex-personrel:jobtitle" qcode="ex-stafftitles:photo">
<name>Staff Photographer</name>
</related>
</creator>
<altId type="ex-idtype:systemRef" environment="ex-acmesys:mdn ex-acmesys:iim">SAD02</altId>
<contributor role="cpprole:descrWriter" qcode="ex-pers:CK" />
<subject type="cpnat:poi" literal="int001" />
<headline xml:lang="en-US">A firefighter walks past the remains of a military
jet that crashed into homes in the University City neighborhood of San Bernardino
</headline>
<description xml:lang="en-US" role="drol:caption">A firefighter in a flameproof
suit walks past the remains of a military jet that crashed into homes in the
University City neighborhood of San Bernardino, California October 18, 2025. The
military F-18 jet crashed on Monday into the California neighborhood after the
pilot ejected, igniting at least one home, officials said.
</description>
<creditline>Acme/John Smith</creditline>
<keyword role="ex-krole:index">us</keyword>
<keyword role="ex-krole:index">military</keyword>
<keyword role="ex-krole:index">aviation</keyword>
<keyword role="ex-krole:index">crash</keyword>
<keyword role="ex-krole:index">fire</keyword>
<slugline>USA-CRASH/MILITARY</slugline>
</contentMeta>
<assert literal="int001">
<POIDetails>
<address>
<line>University City</line>
<locality qcode="ex-city:int002" type="ex-cpnatexp:geoAreaCity">
<name>San Diego</name>
</locality>
<area qcode="ex-state:int003" type="ex-cpnatexp:geoAreaProvState">
<name>California</name>
</area>
<country qcode="iso3166-1a2:US" type="ex-cpnatexp:geoAreaCountry">
<name>United States</name>
</country>
</address>
</POIDetails>
</assert>
<contentSet>
<remoteContent
href="http://www.acmenews.com/content/pictures/hires/20081208/USA-CRASH-MILITARY_001_HI.jpg"
rendition="rnd:highRes" size="3764418" contenttype="image/jpeg" width="2500"
height="1445" colourspace="colsp:USSWOPv2" />
</contentSet>
</newsItem>20.10. Reconciling NewsML-G2 with embedded metadata
Where equivalent properties exist in both NewsML-G2 metadata and embedded (XMP or EXIF) metadata, the IPTC recommends that embedded administrative metadata, such as <located>, MAY take precedence over NewsML-G2 metadata, subject to guidance from the provider, and always with caution.
For example, a picture may have GPS metadata embedded by the camera, which may be different from the <located> property entered by a journalist.
The difference may be one of precision: the GPS co-ordinates may be precise, but hardly useful to the ultimate consumer. Even if a human-readable value is derived from the GPS data, would this be better than the information, if accurate, provided by the journalist? If the difference is due to inaccuracy, the receiver would have no way of knowing whether the journalist has made a mistake, or whether the camera is incorrectly set.