Goals of ITS

• producing a set of tags that people can use to ensure both internationalization and localizability in their schemas - be they based on DTDs, XML Schema or other technologies.
• descriptions of requirements for i18n and l10 which are hard to express via a tag set.
• using additional technologies beyond schema languages to fulfill the requirements.

Requirements for ITS

Identifying date and time

It should be possible to indicate in markup that a string represents an ISO 8601 date and/or time.

Bidirectional text

There has to be a mechanism to indicate the directionality of text.

	Wrong directionality is visualized, relying only on the Unicode bidirectional algorithm
	Feasible directionality is visualized
	Markup to produce the feasible visualization

Requirements for ITS

Identification of content that should not be translated

The content of the trademark element should not be translated.

<trademark its:translate="no" ...

Selection of information with i18n and l10n related characteristics

The information might be textual content of an element. In this case, an element like span could be used. In the case of attribute values, this is a difficult requirement.

Specification of inheritance features and exceptions

Example: All svg elements should not be localized, except the text element in a given context.

Requirements for ITS

Requiring uniqueness of IDs across documents

Linking mechanisms for external information

The error message in this software documentation should be retrieved from an external data base with local specific terms.

<para>If you create a typing error like "strs(s)",
you will get the message
 <xref id="resfile.resx">
  <subst>
   <search>{0}</search>
   <replace><Filename></replace>
  </subst>
 </xref>
.<para>

Overview of further requirements

All requirements are work items on the ITS home page.

• Indicator of Constraints
• CDATA sections
• Handling entities
• Locale / Language identification
• Term identification
• Purpose specification / mapping
• Cultural aspects of the content
• Indicator for metrics
• Limited impact
• Attributes and translatable text
• Localization notes
• Handling of white-spaces
• Multilingual documents
• Annotation markup

Problems we face with the ITS requirements

• Requiring uniqueness across documents is not possible with schema languages.
• Information about attributes is hard to express.
• How to describe linking to external information? The example requires more than could be achived with XLink or XInclude.
• How to integrate ITS in existing schemas without breaking existing validation and further processing chains (e.g. breaking XPath expressions)?
• What to do about versioning?
• How to specify inheritance features?

Our current state:

Schema languages

Alternatives

XML DTDs, XML Schema, RELAX NG; in addition: Schematron?

ITS is - among others - meant for schema authors. The question is: Do we formulate requirements which cannot be fulfilled with all schema languages?

Important criteria:

• Tool support: l10n tools mainly support XML DTDs.
• Expressive power: Do we need a typing system?
• Expressive power: Do we need expressive content models?
• Uniquenesss across documents: Can only be validated via Schematron / XPath 2.0

Namespace sectioning

General idea

Rules describe what namespace should be validated against what schema. According to the rules, documents are seperated into schema-specific element and attribute sections.

Benefits

No need to change the existing schemas. Existing XPath expressions for a given scheme still work for the sections respectively!

Problems

Will this be supported widely? Will this be part of a W3C standard?

Schema annotation

General idea

ITS data categories are realized as additional information to the given schema.

Benefits

Allows for reusing existing schemas, without a need to change them.

Problems

Which annotation format to use? SGML based architectural forms? RDF-based annotations? DSDL part 8? an ITS-specific format?

XML processing model

General idea

The interrelation between processing steps like namespace sectioning, validation, linking to external information etc. are described in a declarative format.

Benefits

Holds everything together :)

Problems

When will it come? There is an urgent need for ITS, and we can't wait too much!

Questions to discuss

• What should be realized on the schema level (i.e. with extensions to schemas or as a schema module)?
• What schema language should we rely on?
• What additional techniques except schema languages should we rely on?
• Should we describe a processing model for ITS?
• Do we need different levels of conformance for ITS, i.e. ITS light (just names for elements and attributes) and ITS full (linking mechanisms, a processing model, ...)?