XLIFF Whitepaper

Introduction

XLIFF is the XML Localisation Interchange File Format designed by a group of software providers, localisation service providers, and localisation tools providers. It is intended to give any software provider a single interchange file format that can be understood by any localisation provider. We investigated many exisiting XML formats such as UIML, OpenTag and TMX and used them as points of reference to aid us in building this new file format.

A number of Information Technology companies set up a working committee to explore ideas around sharing technology or information with a goal of standardising the types of handoffs to we deliver to vendor companies for products we need to be localised.

Although XLIFF is an ongoing initiative that will develop and grow over time, the goal of the initial phase was to create a working specification within a 6 month time span. Although the first phase focused mostly on UI file requirements, the working group endeavored to create a standard that would accommodate support for documentation elements at a later time without having to redesign the basic architecture.

The Mission Statement of the XLIFF working committee is:

Define an extensible interchange data format that is tool independent and standardised for the purpose of localisation and supporting the whole localisation process; this will initially and comprehensively support the collaborator-specific data formats.

Development lifecycle

The working process consisted of the group examining all of the different file types that our

companies use and the processes we create around them and examined what would be required in an XML format to ensure that we could support any localisation process. This XLIFF format had to ensure that we had a mechanism to generate the correct localised files from this XML format as required.

We developed the elements we required and the attributes of those elements and proofed these all along, by developing samples at the same time.

XLIFF elements can be divided into five main categories: the top-level and header elements, the named group elements, the structural elements, the in-line elements, and the delimiter elements. Attributes are shared among them

Why not use an existing format? Why create XLIFF?

We started by examining some existing XML formats such as TMX, OpenTag and some that are under development such as OLIF and ITS. Individually none of these formats fully met our criteria, for the following reasons

·most exisiting formats were too complex

·requirement for a hierarchical, recursive representation of interchange file data

·requirement for a consistent representation of data so that there are not multiple methods to express the same data object

·requirement for a format that could handle variant translations and machine translation

We also examined what we wanted this initial version to handle and what aspects we were deferring to a future phase. Aspects of the first phase XLIFF definition are:

•Project file based

•Multiple files in a single XLIFF file

•Bilingual – single source and target language

•Project information

•Binary objects

•File structure

•Hierarchy shown by <group>

•Source-target pairs in <trans-unit>, resp. <bin-unit>

•On-site localised file creation

•Metadata: versioning, counts, reference material, project data

•Header: <header>

•Skeleton (template) file : <skl>

•Versioning: <phase-group>

•Word counts: <count-group>

•Reference material: <glossary> & <reference>

•Alternate translations: TM, MT, Translator, Reviewer

•Stored next to source text

•Multi‑lingual

•Match-quality information

•Origin

•Translation history maintained

•TM, MT, Translator, Reviewer

•Proprietary tool-specific data

•Named group

•User-definable

•Tool-specific

•Context information: file, group, source, translation

•Named group

•User-definable

•Translator

•TM/MT

•Resources – file should handle the Windows resources as well as other system resources, etc.

•Styles

•Positioning of particular groups/units

•Handles binary data, such as pictures, cursors, sound files, etc.

Proposals for the next phase

Future development proposals include:

•Split/Join files

•SKL Tools

•XSLT for other XML formats

•API definition

•Better TM/MT integration

•Approved translation matches

•Document localisation

Glossary

XLIFF – Xml Localisation Integration File Format

SKL – Skeleton File

TMX – Translation Memory eXchange

References