1	Cost Savings in Combing XML Document Management and Translation Management Software A Case Study
2	What ??? What we look for … Decreasing production costs Publishing on demand What we hear … “XML solves all your problems and increases your revenues” What we get … Conversion to XML is expensive Cost often exceeds $1 Mio Often does not result in expected cost savings
3	Problems that can be resolved by XML Solves publishing problems Single source publishing Multi-channel publishing Solves document quality problems Re-usage of information pieces Document consistency Solves translation / localization problems Layout / format representation Supports workflow better Inherently Internet based
4	Biting the Silver Bullet XML = Software Development ?? Core Comparison with Software Development SE Tools and modern SE approaches solve your problems in software development! It is the “how” not the “what” Structured Analysis Relational databases UML Object Oriented Programming Rapid development … One silver bullet … “Some Fool’s Gold is Silver …”
5	Silver Bullets in Localization SGML & XML Standards like DocBook Editors CMS/DMS Publishing Device independent publishing PDF Print & Digital
6	The Company Humax produces set top boxes in Korea World leader in Digital Satellite Receivers Set-top boxes Audio equipment Some Numbers ~ 100 multilingual manuals per year Different formats like MIF, RTF etc.. 3 technical authors FIGS, Asian languages, Farsi, Arabic (BIDI) Production/translation from English ~ 500.000 $ translation costs ~ 100.000 $ DTP costs (without print) Humax Management considers this unacceptable expensive Search for cost savings
7	Initial Cost Structure in $
8
9	Initial Production Process
10	Ambitious Goals One standardized document format Publishing Simple Editing – similar as in WinWord PDF Output / HTML (for Web Sites) Editor starts the publishing production (Master) On demand publishing Automating Translation Process Translating from Korean (not English!) Integrating translators should be easy Reduce dependency on translation provider Simplified document management > 50 % cost savings
11	Procedure Analysis Where we are right now Recommendations What we need to do to get better Implementation How we do it to achieve results
12	Analysis Variety of document formats WinWord, MIF, various other formats partially language specific encoded Production costs caused by different document formats $ 100.000 Translation Memory databases One external translation provider keeps the tm databases in its office and perceives them as its property Translation system of provider only works out of English Quality of tm database contents extremely diverging Translation process relying only on one supplier Heavy dependency $ 400.000
13	Recommendations Conversion to XML Conversion of all documents to XML Must be introduced consequently Switch source editing to using XML Editor with Style sheets Moving towards internal translation support system Using internally available server solution Converting external databases to standard TM representation Translating from any source language Accessible to different company departments Translators ceasing to work with original documents, working instead with XML (XLIFF) localization standard Publishing into PDF Also into RTF, XHTML, text as well
14	Implementation Procedure Analysis of document structures and databases Semantic structure of documents Necessary format support Mapping of documents into XML DTD design Subset of DocBook -> DocTrans DTD Conversion of existing documents to DocTrans Manual and as much as possible automatically Installation of translation server Conversion of external databases Contains simple document management system Changes in translation are automatically forwarded to tmx DB Several editing tools Installation of a simple Publishing-line Controlled by editors
15	Publishing Tools Documents are managed in a web GUI Collection based XML Editor Creation/Editing of documents Open Source (Xerces, Xerlin, iText …) Uses DocTrans DTD Publisher for Output Generation Open Source (FOP, Batik, JDOM, Xalan, …) Optimized for PDF Technical Author runs it
16	Translation Tools TM („Translation Memory“) MySQL – relational DB TMX – XML Standard for TMs Similarity based search and translation Web based Alignment tool Creation of TMX DB Editor Translation Editor XLIFF based Terminology Tool Web based
17	New Production Process
18	New Cost Structure in $
19
20	Changes …
21	Risks/Problems Printed output differing slightly from pre-conversion output XML ≠ Desktop Publishing Software Bugs/missing support in FOP (Open Source) Conversion of legacy data impossible or extremely complicated Documents must be converted manually Use of TM databases? Lacking Cooperation/Support Technical authors – job loss /reduced income Translation providers – looses influence and revenue by using XML standards Production linkage - partially superfluous as cooperation directly with translators DTP department/provider - becoming abruptly unnecessary
22	Advantages Easier handling of documents Commonality derived from using XML Standards Real in substance derived from single source multiple target and language publishing Solution portable to different OS (Linux) Unlimited re-use of documents Re-Usage in other applications True manifestation of “Knowledge management” Focus of technical authors and translators entirely on text, not on layout Management no longer dependent on translation providers and publishers Improvement in production time and reliability
23	Future Improvements Source document production Using TM technology to improve source document consistency Adding new tools Controlled language MT Grammar Checking Open Source Software …
24	Final What … What we have to do … Conversion must be implemented consequently, no partial solutions Management must enforce conversion process Ability to pushing through towards external suppliers Management must invest time Convince technical authors that new publishing process makes life easier and makes job safer What we have to accept … Be ready to accept partially restrictions in output Not all authors and translators may like it What we will achieve … Cost savings of > 50 % are possible Production times get shorter And the conversion costs are manageable
25	Conclusions