Ford Motor Co. is an example of the MT (Machine Translation) imperative on a grand scale. Ford has manufacturing facilities in Germany, Spain, Belgium, Mexico, and Brazil, where workers assemble vehicles using instructions in their local language. However, all of the instructions are originally created in the United States in English. A single car line can have assembly instructions with as many as 300,000 sentences. Moreover, the instructions undergo frequent changes during the production cycle, requiring quick retranslation and distribution. For such a massive translation problem, MT is the only viable solution.
Ford engineers prepare the assembly instructions using a standardized language. This language has a limited range of syntactic patterns and vocabulary to reduce the possibility of ambiguity. When assembly instructions are prepared, each standard language sentence is stored as a record in an Oracle database. Ford developed its own artificial intelligence system to check the sentences for conformity to the Standard Language rules. When Ford needs to translate the instructions for a particular vehicle and language, the appropriate records are sent to the SYSTRAN MT system, where they are translated using Ford's customized dictionaries and rewritten to the database. The translated instructions can then be sent directly to the PCs of Ford workers at manufacturing sites worldwide. Currently, Ford is producing MTs in four languages: German, Dutch, Spanish, and Brazilian Portuguese. The database, SYSTRAN system, and customized dictionaries are integrated into Ford's Global Study Process Allocation Process (GSPAS), a system for managing labor and manufacturing costs for Ford plants worldwide.
The difficulty of measuring the quality of automatic language translation systems (known as "machine translation" [MT]) has been an obstacle to widespread adoption. With systematic benchmark testing, categorization of errors, and effective dictionary customization, MT technology can yield significant cost and time savings, as well as improved consistency in translations. IDC makes the following observations about the MT market:
- MT quality must be assessed in the context of each user's application. For example, using MT for a chat or instant messaging application is completely different from using MT to translate manufacturing assembly instructions
- SYSTRAN has evolved a process for enhancing MT quality for an individual customer. This process has been validated with actual customers, such as Ford Motor
- SYSTRAN has also developed the SYSTRAN Review Manager (SRM), which helps the customer to manage the MT quality process by allowing them to change vocabulary and linguistic rules. This tool represents an important advance in MT, both technologically and philosophically. Users have never before had the power to modify linguistic rules through an intuitive, interactive process
- By opening up rule modification, SYSTRAN takes a risk, but one that will almost certainly pay off. Engaging users in the process of improving MT is the surest path to increased acceptance and understanding of the technology
In this study
In this IDC study, we discuss the efforts of SYSTRAN to address the issue of machine translation (MT) quality. MT technology faces several important obstacles to broader use in business applications. Most important among them is the issue of quality. Potential users perceive MT quality as uncertain and difficult to improve. All MT implementations require customization, and the scope of this task can be difficult to quantify. Such uncertainties can make ROI calculation difficult or impossible. But despite the perceived intangibility of MT, translation results can be measured and managed effectively. The key is targeted customization of the MT system, with ongoing benchmark testing to ensure that results continue to meet the customer's needs.
To address the quality impasse, SYSTRAN has developed the SYSTRAN Review Manager (SRM), a suite of quality management tools that address the measurement, testing, and customization tasks from a single user-friendly interface. Deployed successfully at Ford Motor Co., the SRM shows us the way to make MT quantifiable and viable for enterprise applications.
situation overview
Intorduction
Despite the growing business imperative to produce and use multilingual business materials, automatic translation technology (known as MT) faces many obstacles to widespread adoption. Three of the most important are:
- Determining return on investment (ROI). There is currently no reliable method for calculating ROI. A useful formula for determining ROI would have to account for a broad range of factors, including language pairs, turnaround time, integration with existing applications, volume, customization, and support
- Evaluation challenges. Translation technology is unfamiliar to most businesses. Comparing among competing products is difficult because it requires substantial knowledge of language technologies as well as a clear understanding of the specific translation requirements. MT products are completely nonstandardized, making feature comparisons difficult and further complicating the difficulty of evaluation
- Uncertainty about quality. The difficulty of assessing and improving translation quality is MT's most intractable problem. Translation quality is inherently subjective and therefore difficult to measure. This is also true of human translation. A text given to three different professional translators will yield three different results. Even if all three are of high quality and accuracy, the subjective nature of human language almost guarantees that there will be differences in interpretation, word choice, and style. It can also be difficult to quantify the effort required to improve translation quality. Tracking ongoing quality levels is also a challenge
The MT Quality Impasse
There is an enduring perception that MT is not yet "good enough" for commercial use. The irony is that MT is better today than it ever has been, and it is in broader use. For many high-volume translation applications, the quality of translation is sufficient to allow understanding of the text. In addition, some MT systems, notably SYSTRAN, have introduced powerful linguistic and integration tools that have increased the user's ability to customize the MT system.
Still, the problem of assuring and measuring MT quality remains a serious concern among potential users. Many of these concerns are well founded, because it is difficult to track and quantify MT quality. There are a variety of factors at the root of the difficulty, which are discussed in the following sections.
Machine Translation Output Is Not Easily Predictable
MT systems work with natural language - a data set that is infinitely varying, ambiguous, and structurally complex. To translate adequately, an MT system must encode knowledge of hundreds of syntactic patterns, variations, and exceptions, as well as relationships among these patterns. It must include ever-changing vocabulary and specific semantic knowledge about the usage patterns of tens of thousands of words. It must accurately identify the parts of speech and grammatical characteristics of words which may, in different contexts, be nouns, verbs, or adjectives, each having many possible translations. Translation also requires a vast store of knowledge about the world, the intent of the communication, and the subject matter.
A human translator prioritizes and selectively applies linguistic rules based on this knowledge. MT software, unless explicitly coded for each possibility, cannot. Thus, MT will never attain the overall quality of human translation. The primary advantages of MT over human translation are speed, cost, and consistency. An MT system gets a great deal more translation done than is possible manually, and MT can deliver translations instantly for time-sensitive content. When a term is entered in an MT dictionary, it will translate it the same way every time, unlike human translators who may choose different translations at different times.
Quality Metrics Depend on the Input Text and the Level of Customization
Potential users want quality metrics that are objective, absolute, and easily compared among competing products. However, translation quality, whether human or software generated, is difficult to quantify. Counting the number of errors in a translated sentence is not revealing because languages do not correspond on a word-for-word basis. An incorrect analysis of one word in the source language, for example, could lead to incorrect translation of several words in the target language. In addition, many errors made by MT systems cause subsequent errors within the sentence. Different systems, and for that matter, different human translators, can produce intelligible, accurate, but different translations of the same sentence. Therefore, for any input sentence, there is no single, ideal output sentence. Finally, some errors are more serious than others, so all errors should not be assigned the same importance.
No Standards Govern MT Systems
Despite the decades of research and development that went into today's MT systems, the industry is still immature. MT systems grew up in very different ways, with many originating with academic research projects or government-funded initiatives. As a result, there are no accepted standards for how MT systems store or process data or what results they produce. Without a standard to measure against, each system vendor is left to make their own claims, which are not directly comparable with the claims of competitors.
Evaluation Is Not Objective
One evaluator might rate a translation as intelligible, while another may not. The judgment of translation understandability is an inherently subjective task that can be affected by factors, including the evaluator's subject knowledge, language facility, reading comprehension, translation experience, and attentiveness.
A Successful Strategy for the MT Quality Impasse
Many potential users give up when faced with the challenges of evaluating, enhancing, and implementing MT. MT vendors recognize the risks, and most have responded by working to improve basic translation quality to increase acceptance. But for most applications, improved translations are not enough. Adopters of MT need comprehensive, easy-to-use tools for measuring the quality of their translations, enhancing dictionaries, and verifying the results. The tools must be accessible to nondevelopers who know the languages and the business terminology for their company. Among the handful of commercial MT systems available today, only SYSTRAN has tackled the quality issue effectively.
SYSTRAN is unarguably the best-known and most comprehensive MT system in the world, having been in continuous development for more than 35 years. SYSTRAN offers 36 language pairs and has the largest dictionaries of any MT system. The company has taken a pragmatic approach, developing a suite of quality measurement and enhancement tools that offer a far more concrete solution to the quality question than esoteric measures of improvements in basic translation quality.
The MT Quality Enhancement Process
Making an MT system work for a particular application is a process, not a quick fix. Improving MT is a cyclic process beginning with review of a translation, update of dictionaries and other linguistic resources, and retranslation to validate the effects. In the SYSTRAN system, the SRM acts as a coordinator, managing access to different customization resources and tracking quality.
Figure 1 - MT QUALITY ENHANCEMENT PROCESS
Source: IDC, 2003
With potentially thousands of dictionary changes, numerous rule modifications, and changing text, it is a challenge to track customization activities and measure results.
The SRM integrates the three steps into a single-process management program with links to the user dictionary, the source and target texts, benchmark files, and interactive translation testing. In addition, the SRM categorizes errors, assigns levels of severity, and keeps track of statistics on the rates of various error types. It can be configured as a Web-based application for single or multiple users. In the latter case, reviewers in different locations can access translations, provide feedback, update dictionaries, and even store their own variant translations for a particular word or phrase. For multinational companies, the SRM allows easy cooperation between sites where different language abilities reside. Some additional benefits of the SRM are:
- Demonstrable method of quantifying MT results
- Increased user autonomy in the enhancement process
- Reduction in the need for continuing customization services from a specialized provider
- Leveraging of the company's own multilingual resources regardless of location
- Increased QA productivity and deeper user engagement in the quality review process
- Improved efficiency in managing translation projects
Step 1: Review Output Using the SRM
During this phase, the SRM functions as an interactive editor, presenting the reviewer with each translation unit in the translated text. The user can modify the translation if it is not acceptable. These modifications are recorded as new entries in the User Dictionary. In the soon-to-be-released version 5.0, the SRM can automatically determine grammatical information, such as part-of-speech and inflection patterns, and enter that information into the new dictionary record for the term or phrase. This function is known as "Intuitive Coding." With Intuitive Coding, people with language and subject knowledge can encode the dictionary without any special expertise in linguistics or programming. The reviewer can also view listings of words that were found in the text, but have not yet been entered in the dictionary. These listings can be entered directly into the User Dictionary from the SRM. The reviewer supplies the translation, and the Intuitive Coding functionality supplies grammatical information.
Step 2: Update Resources and Enhance Source Text
After the review process is complete, the dictionaries are saved, and the document can be retranslated. Reviewers can also open the dictionary records directly and modify or refine the translations or grammatical tags for an entry.
Enhancing the source text is equally important to dictionary building for quality assurance. Translation results tend to be better when the source text is modified to simplify word order and shorten lengthy sentences. SYSTRAN is developing an interactive linguistic tool that allows reviewers to modify the actual translation rules used by the translation engine. Combined with the SRM, the SYSTRAN Translation Workbench is an interactive XML-based editing tool that incorporates the reviewer's changes as rule modifications.
Once it is released, this tool will represent an important advance in MT, both technologically and philosophically. Users have never before had the power to modify linguistic rules through an intuitive, interactive process. Rule access was provided once before in the translation engine developed in the 1990s by Globalink. Code-named "Barcelona," that system was subsequently sold to Lernout & Hauspie and Bowne Global Solutions. The rule language of Barcelona, though powerful, was extremely complex, requiring a great deal of skill in linguistic notation, programming, and languages to use it effectively. In most MT systems, linguistic rules are not even accessible to the user because they are part of the source code.
Perhaps most importantly, the coming release of the SYSTRAN Translation Workbench represents a shift in the attitude of MT developers toward users. MT systems are extremely complex, and developers have always taken pains to protect the user from making naive changes to the system that could have serious consequences for other contexts. This attitude has been a source of frustration to more sophisticated MT users, who eventually reach a wall on quality improvements after building their dictionaries. By opening up rule modification, SYSTRAN takes a risk, but one that will almost certainly pay off. Engaging users in the process of improving MT is the surest path to increased acceptance and understanding of the technology.
Step 3: Retranslate and Validate
Once the changes to the system are saved, the reviewer can retranslate the text to verify that the new entries are in effect. It is important at this stage to check for regressions. Regressions occur commonly in MT output. They can sometimes originate with an incorrectly coded dictionary entry. For example, a user might supply a translation that is correct in the context of one sentence, but incorrect in another context.
The SRM manages regressions with a color-coding system that shows what portions of the text have changed since the last time it was translated. This feature reduces the amount of time spent on reading and comparing the previous translation with the new version by highlighting the areas for focus.
Significance of the SRM
The SRM will benefit SYSTRAN's customers by improving their understanding of translation quality and the process for improving it. The SRM also has broader importance, in that it places far more control over the translation process in the hands of the user than ever before. This may spur changes to the way the MT industry and its customers view each other and lead to more successful implementations of MT.
Case Study: SYSTRAN and Ford Motor
Ford Motor Co. is an example of the MT imperative on a grand scale. Ford has manufacturing facilities in Germany, Spain, Belgium, Mexico, and Brazil, where workers assemble vehicles using instructions in their local language. However, all of the instructions are originally created in the United States in English. A single car line can have assembly instructions with as many as 300,000 sentences. Moreover, the instructions undergo frequent changes during the production cycle, requiring quick retranslation and distribution. For such a massive translation problem, MT is the only viable solution.
Ford engineers prepare the assembly instructions using a standardized language. This language has a limited range of syntactic patterns and vocabulary to reduce the possibility of ambiguity. When assembly instructions are prepared, each standard language sentence is stored as a record in an Oracle database. Ford developed its own artificial intelligence system to check the sentences for conformity to the Standard Language rules. When Ford needs to translate the instructions for a particular vehicle and language, the appropriate records are sent to the SYSTRAN MT system, where they are translated using Ford's customized dictionaries and rewritten to the database. The translated instructions can then be sent directly to the PCs of Ford workers at manufacturing sites worldwide. Currently, Ford is producing MTs in four languages: German, Dutch, Spanish, and Brazilian Portuguese. The database, SYSTRAN system, and customized dictionaries are integrated into Ford's Global Study Process Allocation Process (GSPAS), a system for managing labor and manufacturing costs for Ford plants worldwide.
Unique Challenges
Every MT implementation involves a unique set of customization challenges related to the nature of the text and the intended audience. At Ford, some of these challenges were:
- The texts contain numerous long noun phrases (e.g., insulation assembly body pillar), which must be recorded in the Ford user dictionary to ensure an accurate translation
- All Standard Language sentences are written in Imperative form. Declarative sentences are the most prevalent type in most English texts, so grammatical coverage of imperatives tends to be less robust
- Standard Language uses modification rules that are different from the rules for English. Modifying words can be placed after a noun, instead of before it. For example, the phrase "body panel large" is allowable in Standard Language, even though it is grammatically incorrect in English
- Ford uses a specialized vocabulary. Some of the vocabulary is common to automotive manufacturing in general, but some terms can be specific to the specific plant or manufacturing team. Standard language contains 2,500 Ford-specific terms, 13,000 noun phrases, and over 1,000 abbreviations and acronyms. Ford uses an artificial intelligence system to review its assembly instructions and ensure they conform to the Standard Language rules
- Spelling variants are common. The acronym for "antilock brakes" may be written as either ABS or A.B.S
- Writers can insert free-form comments that do not conform to the Standard Language rules
- Ford's bilingual engineers do not have the time to review translation results
- Standard Language is usually written with no punctuation. MT systems are sentence based, and they rely on proper punctuation to help segment sentences, clauses, and lists
- Standard Language is always evolving. The MT system and its dictionaries need updating to account for the changes in Standard Language
MT Integration at Ford
Ford and SYSTRAN collaborated successfully to address these challenges, integrating SYSTRAN into the GSPAS system in 1998. Today, the system is in use at Ford's worldwide manufacturing plants.
SYSTRAN analyzed Ford's texts to identify frequently occurring technical terminology and built a custom dictionary for the application. It was also necessary to map abbreviations to full words (e.g., [ASSY - ASSEMBLY]). SYSTRAN modified its translation system to account for modifiers that occur after the noun. Dictionary development is only one part of linguistic customization. To customize the rules of the translation system, SYSTRAN uses an XML-based "style sheet" that allows users to select from configurable rule categories. The categories of errors can be tabulated after the review is complete, offering insight into the nature and frequency of problems in the translation.
After initial tests, it was clear that some preprocessing of the assembly instructions would help translation quality, especially with embedded free-form comments and titles, neither of which conform to the Standard Language syntax. In addition, inserting articles (e.g., the) before nouns would help the MT system to identify the correct part of speech. For some languages, these problems are being addressed by automatically preprocessing the text prior to translation.
Ford also identified dialect and text size differences as important areas for quality enhancement. Many languages have variant dialects, though the differences in speech are usually far more extensive than in written English. For example, a coastal Maine resident and someone from the deep South might have difficulty understanding each other's speech. But in written form, their language is very much the same. The same principle applies with translation. In Spanish especially, there are numerous dialects. Although the differences are more prominent in speech than in writing, there are nonetheless some terminology issues among Spanish dialects. The quantity of text for any given message varies depending on the languages involved. For English to Spanish translation, for example, translations are generally 15-20% longer in Spanish than in English. This has implications for how the text is displayed and the size of the text window in the user's application.
Future Outlook
The use of SYSTRAN has helped Ford to translate its large volumes of assembly instructions into four languages. More than 1 million records have been translated. Ford has been able to deliver an accuracy rate of 90% for English/German translations. Ford deployed a Web-based customer dictionary tool in 2002 that allows engineers to introduce new dictionary entries and corrections to translation errors. Modifications to the Standard Language have been introduced as a result of translation feedback.
Essential Guidance
The MT quality impasse can be overcome with customization, ongoing error tracking, and testing against benchmark files. This process can be intimidating to new users who are unfamiliar with MT technology. To deploy MT successfully, the vendors must provide guidance and support to the user until sufficient knowledge is built up within the organization to manage translation quality independently.
When it is effectively customized and tested, MT does produce cost and time savings. Ridding potential users of the notion that MT is a "plug and play" solution is perhaps the MT industry's most important educational objective. The SYSTRAN implementation at Ford Motor provides an excellent case study of how MT, when properly customized, can solve a critical, large-scale translation problem. Other MT users and vendors would do well to follow Ford's example.
Learn more
- Machine Translation Engines: An Evaluation of Output Quality (IDC # 22722 , June 2000)
- MT and TM: ESTeam's Winning Solution (IDC # 28367 , November 2002)
- Worldwide Globalization/Translation Software Market, 2002 (IDC # 28166 , November 2002)
- SYSTRAN and the Reinvention of MT (IDC # 26459 , January 2002)
TRANSLATION SOFTWARE
- CASE STUDIES / ACTUAL CUSTOMER TESTIMONIES
Symantec
is a global leader in infrastructure software, enabling businesses
and consumers to have confidence in a connected world. The company
helps customers protect their infrastructure, information, and
interactions by delivering software and services that address
risks to security, availability, compliance, and performance.
Headquartered in Cupertino, California, Symantec has operations
in 40 countries.
Within Idiom WorldServer, the Fuzzy Score is set here to 75%. SYSTRAN
automatically translates segments lower than this threshold. Translators
review coloured segments within the Browser Workbench.
« Our main objective is to offer quality product documentation
and technical support to our international customers in a cost effective
manner. With the roll-out of our global content management system,
we have control over our documentation production process. Our metrics
show a 100% increase in word counts managed by our internal translation
team since the introduction of SYSTRAN technologies into our localization
process. Automated translation software is a powerful innovation
when integrated into a streamlined translation workflow. It clearly
boosts our globalization capacities. »
Fred Hollowood & Orla Clifford,
Global Language Services, Symantec.
| Challenges |
Solutions |
Results |
- International customers to service with a variety of content
and formats
- Product documentation published in 20+ languages
- Technical Support content delivered in multiple languages
- To help and support international customers in their own
languages
|
- Integrate SYSTRAN Translation Technology with an online
global content management workflow system
- Automatically translate with SYSTRAN no match segments
in translation memories
- Product Documentation and Technical Support content in
English, translated in French, German and Japanese
- Chinese, Spanish and Italian localisation as a second
step
|
- Personal productivity for internal translation team doubled
- A cost effective solution which makes it possible to increase
volume of translated content
- Address translation projects with scalability and cost
control
- Control terminology and streamline methodology to smoothly
enable new language pairs
|
What are the main reasons that invoked
your interest in machine translation for use within your Global
Language Services department?
Symantec operates in more than 40 countries. As a leader in IT infrastructure,
we deliver sophisticated products and services to international customers.
As an international company, we focus on globalization early in the process.
Our Global Language Services division addresses the needs of localized
product documentation and technical support content. With multiple acquisitions
completed in the last number of years, Symantec’s globalization
requirements are complex and diverse. Faced with these increasing needs,
we chose to evaluate a rapid, cost-effective translation technology.
Our goal is to reduce translation costs to 35% and reduce translation
time by 66%, while maintaining quality standards.
What are the main reasons that invoked
your interest in machine translation for use within your Global
Language Services department?
Symantec operates in more than 40 countries. As a leader in IT infrastructure,
we deliver sophisticated products and services to international customers.
As an international company, we focus on globalization early in the process.
Our Global Language Services division addresses the needs of localized
product documentation and technical support content. With multiple acquisitions
completed in the last number of years, Symantec’s globalization
requirements are complex and diverse. Faced with these increasing needs,
we chose to evaluate a rapid, cost-effective translation technology.
Our goal is to reduce translation costs to 35% and reduce translation
time by 66%, while maintaining quality standards.
Can you describe the first steps of your machine translation
project?
We started this project by integrating SYSTRAN technology with Trados
Translation Memory technology. Our internal linguistic teams quickly
gained confidence in maintaining customer specific dictionaries and post
editing. Indeed they doubled their personal throughput of translation
content provision over the first six months of the project. After this
initial success, we integrated MT with Idiom’s WorldServer system
and applied this model to all of our workflow enabled translation resources.
How did you choose SYSTRAN among the other competitors?
We chose SYSTRAN from among a range of MT technology suppliers for the
following reasons:
- They provided a comprehensive language spread for global deployment
.
- The rules based approach allowed us to begin translation without
significant engine training
- Their client-server based solution allowed centralisation of
customer specific dictionaries in addition to general scalability.
- The robust well documented API allows us to integrate this technology
with other third party language technology tools.
- The customisability of the solution has allowed us to quickly
tailor the technology and process to our various requirements.
Can you describe your customised solution for product documentation?
The basic idea is to use machine translation in tandem with a computer-aided
translation tool. We never intended to ‘replace’human
translation but instead wanted to assist professional translators
with innovative tools. We developed an application to link machine
translation processes between SYSTRAN and Trados. We called it SymGlue
(i.e. Symantec Global Language User Environment).
In short the process leverages existing translations from Trados
TM’s and then exports lower value or no match segments to a
TMX file.
We then pass this file to SYSTRAN for machine translation and finally
re-import the machine translated TMX file back into the Product Translation
Memory. We have used this process on a variety of content including
Product Documentation and Technical Support in a variety of languages,
specifically Japanese, Simplified Chinese, French, German, Italian
and Spanish. Underpinning this solution is a robust and well embedded
practice among our Infodev writers of authoring content in a controlled
way. We have deployed controlled language checking software (acrocheck)
to help our writing teams in this practice.
What were your results once you introduced machine translation into
your translation process?
In a very short time, the throughput of our internal translation team
doubled from an average translation throughput of 1500 words per day
to 3000 words. When applying this process to larger documentation sets
we have aimed and achieved an increase in throughput of up to 60% in
some languages.
With the roll-out of Idiom WorldServer in 2009 and the SYSTRAN integration
with same, we expect to further increase our productivity. The integration
of SYSTRAN and Idiom WorldServer give us a completely automated translation
process. We gain control on delivery timing and have the potential
to deliver ever increasing amounts of previously untranslated content.
How do SYSTRAN and Idiom WorldServer work together?
Both companies have extensive API’s which facilitated their
integration in our process. Future innovations in the development
of both products will allow for an ever tighter, more seamless integration.
What are the points you appreciate in SYSTRAN translation
technology?
Apart the overall good translation quality for the off-the-shelf
engine, we appreciate the powerful customization capabilities within
the SYSTRAN modules. With SYSTRAN Translation Style Sheets, we were
able to manage Framemaker and XML specific tags within the product
documentation content, neutralising the potential issue that they
present to a MT engine. We are now starting to use this XML-aware
technology to provide context sensitive translation based on XML
tags or Framemaker elements. We are also interested by the new metrics
proposed in SYSTRAN version 6, in particular the complexity and the
confidence scores to have more control on the translation reliability
and accuracy.
EADS
is a global leader in aerospace, defence and related services.
The Group includes the aircraft manufacturer Airbus, Eurocopter,
the world's largest helicopter supplier, EADS Astrium, the European
leader in space programmes from Ariane to Galileo and MBDA, the
international leader in missile systems. EADS largely deployed
SYSTRAN on its Intranet. MBDA ensures the continuous enrichment
of the translation tool.
"The dictionary coding at the heart of machine translation
removes ambiguities so at MBDA, we load a number of noun phrases
and verb phrases into SYSTRAN to maximize the coverage rate and enhance
translation quality. The higher quantity of expressions contained
in the user dictionaries leads to improved translations. The various
terms and phrases are initially created by MBDA and then transferred
to the SYSTRAN Translation Server on the EADS portal, which is accessible
by all EADS users. Even basic machine translations are now of a good
quality, meaning the number of requests for expensive human translation
services have been reduced. Users are increasingly going to the self-service
portal."
Michael Hoff and Eliane Grosheny-Langlois,
Translation Department, MBDA.
| Challenges |
Solutions |
Results |
- Improve communication and knowledge share between EADS
sites located in the UK, France, Spain, Italy and Germany.
- Cover all languages spoken within the organisations, including
English, French and German at first, then Italian and Spanish
as a second phase.
- Reduce requests for human translation, and associated costs.
- Assist organisational changes and enable employees to
work in a multilingual environment.
|
- SYSTRAN Enterprise Server installed within the firewall
provides a translation service inside the EADS portal.
- MBDA Translation Department builds large dictionaries in
defence, aerospace, engineering, management and legal from
translation samples and by using terminology extraction tools.
- MBDA Translation Department offers machine translations
combined with a quick post-edition as a free-of-charge service
which provides enhanced translations against inhouse requests.
|
- High level translation quality with an out-of-the box EADS
translator obtained from customised dictionaries.
- Strong, and increasing, use of translation requests inside
the EADS ESIS portal - with more than 30,000 translations
per day.
- Direct cost savings of 30% of the yearly budget on human
translation services at MBDA; combined with advantages of
a close-to-real-time service.
|
A few minutes with the MBDA Translation Department…
With an integrated framework covering France, the UK, Italy and
Germany, MBDA is ideally positioned to offer tailored defence systems
that address the requirements of its defence and domestic customers
in Europe. Now, with employees working across all these countries,
MBDA is facing an increased need to knowledge-share. Tools are required
to enhance collaboration between entities and to reduce language
barriers between the employee base.
MBDA must handle effectively a high level of translation requests
from engineering, management, product development, quality and legal
departments so started to look for, and assess, machine translation
solutions as an innovative initiative under the sponsorship of the
EADS Group.
The project originally started in 1999 with MBDA acting as the project
leader and main provider of terminology for customisation. EADS then
performed a direct comparison between two language translation technologies.
At that time, SYSTRAN - available in version 3 –boasted the
higher level of translation quality and was therefore the solution
that was chosen.
The basic idea was to create an overall translation service shared
by all entities within the EADS Group. Three language pairs were
selected: French-English, German-English and German-French.
During 2000 and 2001, the project team spent most of its time collecting
various terms from many different entities. MBDA played a central
role in gathering all vocabularies, restructuring them and loading
them into the machine translation system.
EADS and MBDA then undertook a major campaign to adapt the terminologies
to suit the other translation software which was then in use. "It
was painful because using the other translation software meant we
had to manually code all inflections of each term."
At that time, the machine translation service was then made available
to end users through a single page on the EADS Intranet. The service
became popular very quickly, so in 2001, costs of terminology integration
started to become more and more important.
At that time, Michael Hoff produced an average of 120 entries per
day, integrating them into the dictionaries of the translation software
then in use. To increase the integration flow, EADS/MBDA started
to subcontract additional professional services but the model was
not scalable and became expensive as the requirement was to integrate
more terms and expressions into the system.
During 2003, pushed by EADS growth and the integration of entities,
the company wanted to extend the system and add Italian. At the same
time, SYSTRAN released its new version product which featured Italian
language pairs with both English and French.
Following a re-assessment of SYSTRAN, the company then decided to
implement SYSTRAN as the preferred EADS Translation Server solution.
According to Michael Hoff "The IntuitiveCoding technology enabled
a large increase in productivity. At the end of 2003, we had a backlog
of around 100,000 terms to be coded. With the old translation software,
this task was impossible to complete. With SYSTRAN and its Dictionary
Manager functionality, we loaded the data within a few weeks."
Compelling business reasons to use machine translations are low
costs and quick results. Often, users have time constraints and need
to translate a large volume of text quickly: such as operational
documents, design specifications, engineering documents, quality
plans, meeting minutes, good manufacturing practices, management
documents, etc.
In early 2004, the MBDA Translation Department (at that time part
of the Quality Directorate) decided to begin work on French-English
machine translation on a wide scale through bulk integration of the
terminology along with the corresponding validation work in corpus
documents. At this time, the EADS level of activity was at approximately
30,000 translations per day.
On the EADS translation server, English to French is the most popular
language pair (31%), followed by English to German (22%) then French
to English (19%) and German to English (14%).
Along with the EADS translation service portal page, an email-based
user comment service was also developed. This service saw a further
increase in quality using a dialogue method across the users - taking
their remarks into account and thus providing the human element of
the process.
The MBDA Translation Department also then noticed that comments
subsequently started to decrease, whilst the number of users and
translation requests increased.
In addition to this successful translation service portal initiative,
MBDA was also working to reduce the costs of human translations.
The MBDA Translation Department completed more than 10,000 translation
requests in 10 years with its classic human-based professional translation
service. To address these translation demands, MBDA had defined an
intermediate state which was called "enhanced machine translation".
It offered to quickly post-edit the output of SYSTRAN translation
and produce a translation quality that was close to professional
translation.
The mission of the MBDA Translation Department now is to provide
all kind of translations - offering three classes: classic-professional
translations done by humans, machine translations performed using
computer technologies, and enhanced machine translations performed
by machine translation software with minimal human post-editing.
"The three types of translations fit perfectly with all the
translation needs that we are faced with inside our organisation," Hoff
added. "Around 1,000 human translation requests were addressed
to the department last year, 20% of which were processed using the
new enhanced workflow which reduces the overall demand for expensive
human interactions. Between 2004 and 2009, we saved around 30% of
our yearly budget, thanks to the new service.
The SYSTRAN system was customised with large EADS and MBDA dictionaries
to offer the capability of producing a translation that was up to
90% complete. Revisions then only took a few minutes per page.
Quality output is mainly dependant on the effectiveness of the user
dictionaries," Hoff concluded.
An ideal machine translation solution would feature complete disambiguation
of all ambiguous categories (semantic and context) in the source
document.
To obtain the quality approach, the translation solution creates
dictionaries that contain as many noun collocations, multi-purpose
expressions and verb phrases as possible.
"The dictionary coding at the heart of the translation machine
removes ambiguities so at MBDA, we load a number of noun phrases
and verb phrases into SYSTRAN to enhance translation quality. The
higher quantity of expressions contained in the user dictionaries
leads to improved translations."
SYSTRAN offered a key advantage against the previous translation
software - IntuitiveCoding technology. SYSTRAN automatically codes
all inflections for a given term and, in addition, offers a very
powerful coding technique that enables context-specific translations
of the terms used.
MBDA is now using specific tools to automatically extract the expression
candidates from the texts. Then decides whether to code them or not
within the dictionaries.
Thanks to the SYSTRAN dictionary prioritisation sequence - which
is formed by subsequent layers of dictionaries - the most recent
term additions are used before older terms. Effectively, the system
evolves to produce increased quality translations.
The architecture is also error tolerant as errors can be corrected
in the uppermost dictionary without requiring corrections in the
lower layers. It also reduces the workload for ensuring the system
remains operational.
At present, SYSTRAN handles more than 300,000 terms within the user
dictionaries which are organised in six user domains: Aerospace,
Defence, Information Technology, Legal & Business and Optronics.
These customised resources are then periodically transferred to EADS
to also enhance the translation services on the Intranet portal.
With a highly customized translation system, EADS portal users can
now easily select from a list the right user domain, and then enter
the text or upload a document for translation. Supported formats
are DOC, PDF, TXT, RTF and HTML.
Usage rate is also growing, demonstrating that it is a useful business
tool that can empower companies to reduce language barriers throughout
their organizations.
Typical impact density of EADS / MBDA user dictionary terms
in a machine translation (without any manual post-editing)
| Communiqué de Presse MBDA (mai 2009) |
MBDA press release (may
2009) |
| ASTER |
ASTER |
| De l'autoprotection à la défense anti- missiles
balistiques |
From self-protection to anti-ballistic
missile defence |
| Le programme Aster est le fruit des analyses opérationnelles
et techniques menées par la France et confortées
par les études de l'OTAN, puis de la coopération étroite
et volontariste de la France, l'Italie et le Royaume-Uni, pour
se doter d'une famille de systèmes anti-aérien
et anti-missile terrestre et naval. Ainsi, six armées
de trois pays européens financent le développement
de variantes navales (Aster 15 et Aster 30) et terrestre (Aster
30) de moyenne portée. Au plan industriel, Aster est aussi
un programme structurant de l'industrie européenne de
défense avec plus de 1700 missiles déjà commandés. |
The Aster programme is the fruit
of the operational and technical
analyses carried out by France and
consolidated by the NATO studies,
then of the close and voluntarist co-operation of France, Italy
and the United Kingdom, to
procure of a family of ground-based
and naval anti-aircraft and anti-missile systems. Thus,
six armies of three European countries finance
the development of naval variants (Aster
15 and Aster 30) and land (Aster
30) medium-range. At
industry level, Aster is also a structuring programme of
the European defence industry with more
than 1700 missiles already ordered. |
| Les systèmes à base d'Aster bénéficient
de la technologie la plus aboutie pour traiter les différentes
menaces et notamment la menace balistique de courte portée.
Anti-missile de conception, les choix technologiques lui confèrent
une très grande manœuvrabilité et une très
grande agilité ce qui lui permet d'intercepter dans la
majorité des cas par impact direct des cibles avec des
vitesses de rapprochement de l'ordre de 4000 km/heure. |
The Aster-based systems profit
from the cutting-edge technology to treat the
various threats and in particular the short-range ballistic
threat. Anti-missile by design,
the technological choices confer
a very great manoeuvrability and
very an high agility to him what
enables him to intercept in most cases by direct
hit of the targets with closing
velocities of about 4000 km/h. |
| La manœuvrabilité d'un vecteur étant inversement
proportionnelle à son poids, Aster est donc aussi petit
et léger que possible. La conséquence directe a été d'aboutir à une
dichotomie entre le volume d'interception, dépendant de
la propulsion principale et l'efficacité à l'intérieur
de ce volume. L'Aster est donc placé dans son domaine
d'interception par un accélérateur largable, dimensionné en
fonction des besoins opérationnels. Pour satisfaire à l'exigence
de couverture sur 360° dans une fenêtre de tir réduite
et avec des portées d'interception courtes, le lancement
de l'Aster est vertical et les tuyères de son accélérateur
sont flexibles. Elles sont dérivées des technologies
spatiales (Ariane). |
The manoeuvrability of a vector
being inversely proportionate
with its weight, Aster is thus as small and light as possible.
The direct consequence was to
lead to a dichotomy between the interception
envelope, depend on the main propulsion and
the effectiveness inside this volume. The Aster is thus placed
in its interception envelope by
a jettisonable booster, dimensioned
according to the operational requirements. To
meet the requirement of 360° coverage in a reduced
firing window and with short interception
ranges, the launch of the
Aster is vertical and the nozzles of
its booster are flexible.
They are derived from space technologies
(ARIANE). |
| Le choix d'un pilotage aérodynamique fort conjugué à un
pilotage pyrotechnique additionnel appliqué au centre
de gravité du missile intercepteur compensant le temps
de réponse du pilotage aérodynamique a permis de
concevoir un vecteur à l'agilité et à la
manœuvrabilité inégalées dans tout
son domaine d'action et ce principalement à haute altitude,
ce qui le distingue nettement de ses concurrents. |
The choice of a strong
aerodynamic flight control combined to an additional pyrotechnical
flight control applied to the centre-of-gravity of
the intercepting missile compensating
for the response time of the aerodynamic
flight control made it possible to design a vector with
the agility and the manoeuvrability unequalled
in all its operation area and
this mainly at high altitude,
which clearly distinguishes it from its competitors. |
MBDA and EADS dictionaries offer extensive coverage on any corporate
text or document. As a consequence, SYSTRAN
out-of-the-box translations are always of a high
quality and require minimal, or no, revisions.
TESTIMONIAL
BY: James Mentele, Information Scientist with Dow Corning Corporation
had these comments about Systran Software.
Accuracy - "Systran
is superior to other translation systems that we have examined."
Overall
Productivity - "In our experience, Systran
is superior to other translation software. It is excellent for
assimilation and can be a great aid to a less-than-fluent multilingual
person. We believe that we will have a distinct competitive advantage
if we can enable workers to produce work-products as fast, safely
and accurately as possible (usually in the person's native tongue).
but tie that activity into global projects and workflow for greatest
leverage of impact."
Completeness - "Systran
excels compared to other software tools that we have investigated
- both in breadth of specialized dictionaries, as well as features
like proper nouns (name) extraction, .nfw (not found word) file to
highlight potential problems, romanji transliteration of English
and katakana/hiragana of Japanese."
Cost Savings - "The
seconds of machine time per page with Systran is dramatically less
than the hours per page of a human translator. It doesn't take many
pages to recover the cost of the PC and the Systran Software."
Manpower Savings -
A major strategy of a global company is to achieve improved results
with fewer people-hours by being able to leverage specialized skills
and knowledge worldwide. Such a goal is not realistic without the
ability to translate work products. Translations of such volumes
and subject are not realistic without large numbers of multilingual
domain experts. Systrans's specialized dictionaries greatly reduce
the need for such rare capabilities. Our studies have shown that
the time spent by employees translating reports in 1993 was equivalent
to salary and benefits of the $6 million in Japan alone."
Time Savings - "The
biggest time savings occurs for employees who must translate their
work products (monthly reports, project status reports, etc.) into
another language (primarily English) for management."
TESTIMONIAL
BY: The National Air Intelligence Center's 28-Year Relationship
with SYSTRAN Software, Inc. A Case Study
Agency Description - The
National Air Intelligence Center (NAIC) at Wright-Patterson Air Force
Base, Ohio, is the Air Force's single all-source aerospace intelligence
center. Its mission is to support the war fighter, the acquisition
community and the national policy maker by acquiring, collecting,
analyzing, producing and disseminating foreign aerospace intelligence
to the U.S. Air Force, the unified commands, sister services, other
members of the intelligence community and allies.
The NAIC was formed in 1993 by combining the Foreign Aerospace Science
Technology Center, which focuses primarily on the production of scientific
and technical intelligence, and the 480th Intelligence Group, which
focuses on the preparation of cockpit-oriented target material and
mission planning intelligence. In 1994 the 497th Intelligence Group
Directorate of Assessments, which provides analysis support directly
to the Air Staff and other intelligence organizations in the Washington,
D.C. area, was integrated into NAIC. The consolidation of these three
units resulted in a signification mission expansion enabling NAIC
to provide fully integrated intelligence products tailored to customer
requirements. The NAIC has a staff of approximately 1,700 people.
SYSTRAN's Involvement with the NAIC In
1968 SYSTRAN Software Inc. was contracted to develop Russian-to-English
machine translation for the U.S. Air Force through a predecessor
agency of NAIC. In 1969 the first SYSTRAN system was tested at Wright-Patterson
Air Force Base. Since 1970, the system has continued to provide translations
for the USAF's Foreign Technology Division. SYSTRAN was used by NASA
during the joint US-USSR Apollo-Soyuz space project in 1974-75.
SYSTRAN translation software is used at more than 30 sites
within the intelligence community. SYSTRAN's Russian-into-English
machine translation program now includes more than a half million
words and operates at more than 90 percent accuracy on technical
texts.
Translation Needs - The
National Air Intelligence Center has extensive translation requirements
in terms of languages and documentation, although a description of
translation requirements is not available.
Currently NAIC translates technical texts, foreign language journal
articles, and systems documentation from nine languages into English
using SYSTRAN Software. As part of its current five-year contract
with NAIC, SYSTRAN will create machine translation systems for several
Eastern European languages, including the first-ever Serbo-Croatian-into-English
machine translation software program.
Translation Strategy - SYSTRAN
Software is used organization-wide for quick information. It is used
by the Translation Services Department for edited and finished translations.
NAIC has a government-wide unclassified network called "NAIC
Open Source Information Service (NAIC OSIS)." This
network uses nine SYSTRAN systems to translate text for Internet
Websites submitted by government users. This is now being migrated
to a secret network and to a top secret network called Interlink.
As a result, SYSTRAN software is accessible on three different
networks from NAIC's Home Page on the World Wide Web for use by
the entire U.S. intelligence community.
Benefits of SYSTRAN Software - "The
most heralded aspect of machine translation from user surveys over
the past 15 years has been time savings." The agency also
notes that "The most heralded aspect of machine translation
from user surveys over the past 15 years has been time savings." The
agency also notes that using SYSTRAN results
in manpower savings, cost savings and an increase in overall productivity
.
TESTIMONIAL
BY: Autodesk
Autodesk Chooses Systran for Multilingual
Customer Support
Autodesk's Evaluation Process: Autodesk
conducted an informal but comprehensive evaluation of MT products
before selecting SYSTRAN for its application. A test suite of representative
technical articles was provided to potential MT providers, and the
results were reviewed by Autodesk's linguists and technical support
staff. The review focused on identifying translation results that
were useful and understandable, despite the stylistic and grammatical
errors that MT systems inevitably produce.
Autodesk followed a software-development approach to evaluating
SYSTRAN. Errors were reported back to the SYSTRAN team. SYSTRAN then
fixed the errors and submitted revised versions of the system to
Autodesk for verification that the changes had been made. This process
allowed Autodesk to observe both the responsiveness of the SYSTRAN
team and the enhancement potential of the technology.
Autodesk cited three factors in the selection of SYSTRAN from a
field of competitors. The first was that SYSTRAN's output quality
reached the threshold of intelligibility that Autodesk felt was needed
to make its deployment successful. The AltaVista deployment of SYSTRAN
for multilingual Web page browsing gave Autodesk confidence in SYSTRAN's
scalability. Less tangible, but equally important, was the impression
that SYSTRAN understood Autodesk's needs better than other providers
and could work with Autodesk to tune the system to its unique texts.
SYSTRAN CEO Dimitris Sabatakakis also believes SYSTRAN's ongoing
extensive development work was an important factor. The system is
undergoing a major revamping of its dictionary structure that will
allow it to leverage its enormous lexical resources more quickly
and efficiently.
Autodesk's Multilingual Customer
Support Application
The Company: Autodesk is the
developer of AutoCAD, a computer-assisted design (CAD) software platform.
Most of the company's applications for specific design requirements,
such as architectural and mechanical design, multimedia, manufacturing,
construction, and geographic information systems, rest on the AutoCAD
platform. Autodesk products are used by more than 4 million design
professionals. Based in San Rafael, California, Autodesk has offices
in 60 other locations worldwide. John Walker founded the company
in 1982. The company will have revenue of approximately $1 billion
in 2001.
Business Rationale: With more
than 60% of its business conducted outside the United States, Autodesk
is a case study for the challenges of information dissemination and
management in a multilingual environment. One of Autodesk's most
pressing challenges is supporting customers across many languages
in a cost-efficient manner. Autodesk provides customer support through
a database of more than 10,000 articles that are accessible from
its Web site. The articles, which average 1,000 words in length,
are written in English only, are highly technical, and are specific
to design issues for various industries.
Unlike the highly dynamic content in chat, email, and message boards
that has been the focus of previous Internet MT applications, Autodesk's
content is relatively stable. Once posted, the text of articles rarely
changes, and only a few hundred new articles are added each month.
The database receives an average of 500,000 hits per business day.
Although the percentage of hits that require translation is unknown,
the potential volume of translation is very large given the size
of the database and the number of users and languages. Frequently
requested articles will be pretranslated and cached to allow instant
delivery to customers and to reduce the load on the translation servers.
Autodesk acknowledges that without MT it would not be able to deliver
multilingual customer support comparable to what it provides to English-speaking
customers. Mirko Plitt, process analyst in Autodesk's Worldwide Localization
department, states that SYSTRAN's "innovative customization
approach was the only answer to our international customers' need
for a multilingual product support knowledge base: translations produced
by general-purpose MT systems are of little use to our non-English-speaking
clients, and a translation workflow involving human intervention
was not a realistic option. The specific machine translation solution
developed by SYSTRAN maximizes the benefit our customers get from
the Product Support Web site and further increases the quality of
service provided by Autodesk.
TESTIMONIAL
BY:Fisher-Rosemount System, Inc.'s Experience
With SYSTRAN Software, Inc. A Case Study
Company Description - - The
Fisher-Rosemount family of companies is the world's largest process
management supplier. Fisher-Rosemount not only leads in many global
market segments, but it also has the industry's broadest process-automation
offering, including process management systems, control valves, regulators,
transmitters, analyzers and related services. Fisher-Rosemount, Inc.,
is a world-leading supplier of process management systems and services.
The company, headquartered in Austin, Texas, maintains key manufacturing
and technology centers in Burnsville, Minn.; La Habra, Calif.; Cambridge,
Ontario (Canada); Leicester, England; and Singapore. Fisher-Rosemount
Systems offers three process management systems, each of which reflects
the company's long heritage as a pacesetter in process automation.
The company, which was established in 1956 as Rosemount Inc., now
has approximately 1,000 employees.
Translation Needs - Fisher-Rosemount
Systems, Inc. has extensive translation requirements in terms of
languages and documentation. Materials that need translation are
sent to SYSTRAN Software, Inc. via E-mail, translated into final,
edited copy at SYSTRAN's world headquarters in La Jolla, and returned
electronically. Fisher-Rosemount requires translation of approximately
400 pages per year.
Translation Software Language Pairs: English German, French,
Russian, Spanish, Chinese and Japanese. Types of Documentation:
Customer manuals for both hardware and software product.
Translation Strategy - - The
Fisher-Rosemount family of companies is the world's largest process
management supplier. Fisher-Rosemount not only leads in many global
market segments, but it also has the industry's broadest process-automation
offering, including process management systems, control valves, regulators,
transmitters, analyzers and related services. Fisher-Rosemount, Inc.,
is a world-leading supplier of process management systems and services.
The company, headquartered in Austin, Texas, maintains key manufacturing
and technology centers in Burnsville, Minn.; La Habra, Calif.; Cambridge,
Ontario (Canada); Leicester, England; and Singapore. Fisher-Rosemount
Systems offers three process management systems, each of which reflects
the company's long heritage as a pacesetter in process automation.
The company, which was established in 1956 as Rosemount Inc., now
has approximately 1,000 employees.
Translation Needs - Fisher-Rosemount
Systems, Inc. has extensive translation requirements in terms of
languages and documentation. Materials that need translation are
sent to SYSTRAN Software, Inc. via E-mail, translated into final,
edited copy at SYSTRAN's world headquarters in La Jolla, and returned
electronically. Fisher-Rosemount requires translation of approximately
400 pages per year.
Language Pairs: English German, French, Russian, Spanish,
Chinese and Japanese. Types of Documentation: Customer manuals
for both hardware and software product.
Translation Strategy - Previously,
translation was handled in various foreign countries on an as-needed
basis. In 1992 Fisher-Rosemount began relying on SYSTRAN for translations
of technical manuals. The relationship has been effective because
of SYSTRAN Software's "speed and cost."
Benefits of SYSTRAN Software Translation
Services - "Having one source for
translation regardless of the language is a great convenience.
SYSTRAN Software, Inc. is very strong
in machine translation. The people we work with at SYSTRAN
are very knowledgeable of computer-aided translation and the
Interleaf program.
It's a great time-saver to receive
the translated files electronically in the proper format. We
receive excellent phone, fax and E-mail response to questions
and problems. We are working with SYSTRAN to develop a methodology
to minimize the costs of revisions within the translated documents.
A weakness in the process is the absence of translation support
for on-line help systems in Windows NT."
TESTIMONIAL
BY:Gaumont Newsreel Archives
SYSTRAN Chosen
to Translate its French Online Newsreel Catalogue for the Global
Marketplace
When Stuart McKay, a freelance film archive researcher,
was looking for historical footage of the First World War for a new
British TV series, he naturally turned to the Web for information
about relevant holdings in French film archives. France, the birthplace
of cinema, has a vast range of holdings of early film in its military
and government archives. But what made Stuart's search particularly
fruitful was the availability of a remarkable online database listing
the complete historical footage held in the Cin�matique Gaumont newsreel
archives. And above all, the ability to search this database in the
universal language of professional media searchers - English.
The Cin�matique Gaumont is a French film library offering
the largest range of French language newsreel and other film holdings
of its kind. Information about the archive is accessible via the
Internet, offering film researchers an unparalleled database for
searching cultural and historical material on celluloid, viewing
excerpts and then ordering them. To render this facility as universally
accessible as possible to researchers like Stuart McKay, Gaumont
contracted SYSTRAN to provide an ondemand translation solution that
would enable researchers to retrieve the database and read its film
descriptions in English. Gaumont has been able to substantially grow
its market for newsreel film archive users by combining the ease
of access afforded by an online database with the communicative effectiveness
of SYSTRAN's Machine Translation technology.
TESTIMONIAL
BY: Price Waterhouse Cooper
Use and Cost Savings: Based
on the success of an initial deployment in Spain, other countries
were added one at a time by customizing the system for each new language
and adding authenticated access for the users in that country. PwC
paid for the service under a global agreement, with individual countries
paying for the cost of local customization (that is, adding PwC-specific
and department specific vocabulary to the generic translation dictionaries).
List price for a machine translation desktop software license is
approximately $1,000, whereas a corporate service, including setup
costs, ranges from $13,500 per annum for 100 users with five language
pairs to $77,200 per annum for unlimited users and five language
pairs.
Results: As with
most deployments of machine translation, PwC found the main benefit
for users to be understanding the gist of documents in a language
they do not speak well. The following uses have been identified to
date.
. A major use is translating the results of Web searches
of internal or external sites. Finding the relevant content and Web
pages prior to translation is a separate and ongoing issue. PwC's
selected search engine does not yet support cross-lingual search
(as is the case with most major search engines), and smaller vendors
with good cross-lingual search were difficult to scale for global
needs.
. Management personnel use the system to get the main
ideas from text that they don't understand (for example, internal
documents, meeting minutes, client documents, e-mail from overseas
clients) to decide whether to have human translation. Frequently,
opportunities can be identified by homing in on document types where
a significant amount of human translation is already performed, but
where understanding the gist is enough for some portion of the documentation
(for example, in a time-critical application such as an investment
opportunity, where the time delays in professionally translating
all relevant documentation would cause missed deadlines).
. Use of the system has allowed specialists to be assigned
to project teams where they may not be skilled in the team's language.
The machine translation system is used to translate agendas, minutes
and project documentation. What helps make this application workable
is knowing the context of the documentation (that is, understanding
what the project is about, knowing the subject matter and knowing
the nature of the document).
. Non-English speakers may use the system to create
a document in English, if they do not know the relevant terminology.
This is used only for internally targeted documents that are not
worth translating professionally. Users become familiar with how
to tailor their original text to help the system work better (for
example, by using simple, unambiguous language and sentence constructions).
In PwC's case, members of the translation department do not use the
machine translation system, because they prefer to use traditional
translation tools. The system is not appropriate for legal documents
(for example, judgments or statutes) or text that may have legal
ramifications. It works well with highly technical documents, but
not where the language is abstract or philosophical. Longer term,
PwC aims to integrate machine translation with its e-mail system
so that users can see the two languages side by side.
Critical Success Factors/Lessons
Learned: Factors that helped this project become successful
included:
. Gaining buy-in from the professional translators
to help evaluate and customize the system, even if they are not the
target users
. Making the business case that this is a knowledge
management tool targeted at individual productivity, rather than
for quantifiable cost savings
. Setting realistic expectations as to the constraints
of the technology; the logon screen has a "health" warning
telling users what the system can do
. Creating a user interface that allows users to access
translation functionality from any application or intranet site without
leaving their documents
. Establishing a global license that makes it straight
forward to add new country sites
. Making the system customizable for each country (and
even individual, if desired). If more budget were available, PwC
would have performed more customization to specific businesses, as
this makes a big difference in the quality of the translation.
Bottom Line: Enterprises
with global operations should examine how machine translation can
improve access to internal documentation and international Web and
intranet sites, as well as enhancing collaboration between employees
in different countries. The technology is not ready for creating
documents for external consumption and is not necessarily a benefit
for professional translators.
Copyright 2002 CS-17-9077,
26 September 2002
See full version of this case study - click
here
TESTIMONIAL
BY: DaimlerChrysler
The Company: The
1998 merger of Daimler-Benz AG with the Chrysler Corporation formed
one of the world's largest automotive companies. The new entity,
DaimlerChrysler (DC) employs more than 372,500 people in 37 countries.
The company's brands include Mercedes-Benz, Chrysler, Jeep, Dodge,
Freightliner, Setra, Smart and Sterling. The DaimlerChrysler Services
division is a leading provider of financial services. The company's
total revenues were $136.1 billion in 2001.
Communication Challenges: With
its large multinational workforce, the newly formed DaimlerChrysler
faced substantial communication challenges in integrating its operations
centers in Germany and the United States. Although the company has
two official languages - German and English - the level of language
skills varies considerably among DaimlerChrysler employees. While
most German speaking DC employees can speak some English, not all
are able to do so with the ease and accuracy that is needed for effective
working relationships. Among the English-speaking staff, very few
have any knowledge of German at all. For both groups, understanding
corporate documentation written in a different language may be difficult
or impossible.
Many of DaimlerChrysler's company-internal documents,
such as human resources materials, are professionally translated
and published for employees. But the merger increased the number
of informal day-to-day communications among employees in the United
States and Germany.
These interactions, which include email messages, internal
Web pages containing message boards or corporate documentation, and
other unpublished company texts, were difficult for employees with
limited language skills. Traditional human translation was not a
viable solution because of the volume, transience and immediate delivery
requirements of informal communications. In addition, human translation
would be prohibitively costly.
Evaluating MT Solutions: As
U.S. and German interactions increased following the merger, DaimlerChrysler's
Language Services Department began to receive numerous requests for
automated translation support. With the popular success of machine
translation applications such as AltaVista, many DC employees had
witnessed firsthand the benefits, as well as the potential pitfalls
of machine translation software. In response to the many inquiries,
Edith Kroupa, DC's manager of language technology implementation,
organized an evaluation of machine translation solutions and their
ability to meet DaimlerChrysler's unique requirements. Four commercial
MT systems participated in the evaluation. The evaluation entailed
building a profile of DaimlerChrysler's requirements, identifying
the features of the MT systems, and comparing the fit between the
two.
The chief considerations for
DaimlerChrysler were:
! German-English bidirectional language pairs
! No installation of client software
! Seamless integration with DC's IT environment
! Low performance costs
! No maintenance requirements
! Ease of use and access
With thousands of employees, DaimlerChrysler recognized that installation
of client software would create enormous maintenance burdens for IT staff.
As a result, a centralized server installation, and integration with
DC's IT environment was essential. Low performance costs were also important
to achieving a return on investment because the potential volume of translation
was very large. Since very few DaimlerChrysler employees were familiar
with translation technology, ease of use was also an important consideration.
The company also evaluated the quality of translation among the four
systems using a combination of published studies and internal testing
of DC documents. Of the four systems, SYSTRAN met best DaimlerChrysler's
requirements for ease of integration, low performance costs, language
pairs and translation quality.
The current production system is used by 25,000 DaimlerChrysler
employees for the translation of Web pages, emails and corporate
documents. Users access SYSTRAN using a browser-based interface that
interacts with a central SYSTRAN intranet server located within DaimlerChrysler.
The terminology dictionaries used by SYSTRAN are maintained by Language
Services to ensure consistent terminology usage and complete coverage
of DaimlerChrysler vocabulary. Language Services also operates a
help desk for SYSTRAN users. SYSTRAN's staff has worked closely with
the Language Services group to provide technical support and customization
during the implementation phase.
Results and Future Plans: DaimlerChrysler
has seen an increase in the productivity
and effectiveness of informal business communications through the
use of SYSTRAN. The production system currently processes
more than 4,000 translation requests each week. DaimlerChrysler conducted
user acceptance studies during the implementation process. The results
showed that although users recognized the limitations of non-customized
machine translation, they still found it to be a useful tool for
translating informal communications. Based on the success of the
implementation, Daimler Chrysler is preparing to launch SYSTRAN machine
translation capabilities on the company's employee portal. This will
extend the reach of the technology to a broader range of users, and
a wider variety of document types. Additional language pairs will
be deployed based on the needs of users.
The DaimlerChrysler MT story is unique because the
initiative to deploy machine translation originated with employees,
not company management. Historically, management-driven imperatives
to deploy MT within corporations have met with limited acceptance,
especially within corporate translation departments. The initiative
of DC employees in the choice to use MT has been instrumental to
its success. It also illustrates the powerful impact that Internet
MT applications have had for individuals, who in turn can evangelize
the benefits of MT technology within their companies. This grassroots
motive for MT deployment may be a bellwether for corporate implementations
of the future.
If you have used Systran Translation Software and would like
to share your comments with others please send us an email.
Buy and 
