Main Content

Automated Translation Technology
Review, Business Applications & Implementation

Instructors: Kirti Vashee and Heidi Depraetere

09:00 - 17:00 : Thursday, December 11

Overview

Following 50 years of research and development, machine translation (MT) is now reaching acceptance even in the corporate world with localization and customer support professionals. The most promising advances in translation automation are being made with statistical approaches, and Google, Microsoft, Asia Online and IBM continue to advance the state of the technology. Furthermore, open source SMT (Statistical Machine Translation) initiatives like the Moses Project make it easy for all companies to explore the use of this technology.

Translation memories, glossaries and terminology databases, accumulated over more than a decade of localization activities, represent a new value as training data for SMT systems. Statistical machine translation is making serious inroads in everyday applications such as search, multilingual knowledge base support and high volume localization projects. The list of SMT providers includes big names such as Microsoft, IBM and Google but also include dedicated technology providers such as Asia Online and Language Weaver who make it possible for Fortune 500 companies to now use automated translation to rapidly convert large and growing knowledge content into multiple languages.

Today, global corporations see that the new dialogue with the global customer base is increasingly conducted through the web. It is becoming increasingly important for corporations to engage with customer on this platform and automated translation offers a solution for this. Localization teams can lead the way and guide companies in the best way to do this, providing the critical linguistic support necessary to make these initiatives successful. SMT technology has evolved to a point that linguists working with this technology can provide excellent results on what were once considered impossible translation projects.

We will also point out where the challenges are the greatest and where the industry will need further development to enable successful use of this technology.

In this full-day workshop we provide a clear overview of the underlying theory of SMT and other approaches, determine where it makes most sense, point to best practices and also show some examples of major successes in global corporate applications.

The workshop will help to answer questions like:

  • Why is this so different from the traditional Rule-based systems (RbMT) of the past? Why does this approach show more promise if RbMT has not really been able to show meaningful progress in 30 years?
  • What is driving the increased interest and adoption of automated translation?
  • Are the promises of SMT true and applicable to the localization industry?
  • What are the basic working principles of SMT and what do I have to do to ensure the best results?
  • What kinds of skills are necessary to be successful with SMT?
  • How much data do we need to train SMT? Is it true, the more data the better? How do we ever get enough clean and trustworthy data? What does clean data look like?
  • Is regular TM useful for training and what are best practices on optimizing data for SMT training?
  • How do we measure the quality and determine what is acceptable to the end-user? What are useful metrics to determine if an automated translation project is successful and how can a pilot project be sold to top management?
  • What are best practices in post-editing SMT output? How is post-editing SMT different from post-editing output from rule-based engines? How do we ensure that SMT systems learn quickly and improve rapidly?
  • Can the SMT system learn from corrections made during post-editing? How do we customize SMT and improve quality on an ongoing basis? Does SMT change the nature of translator’s jobs? What kinds of linguistic skills would be most useful in the SMT world?
  • How do we integrate SMT in a typical localization workflow and translation management system?
  • What is the cost of deploying SMT? Is it true that we need hundreds of computers?
  • How can we engage the community in helping with SMT post editing and translation system cleanup? How do SMT and crowdsourcing connect? Can you really use the crowd on corporate content? How do you maintain quality when you do this?
  • Will data sharing associations enable more rapid progress with SMT? Why or why not?

Who Will Benefit from this Workshop?

Directors/VPs of Marketing, Advertising, Engineering, Manufacturing, International Sales, Operations, IT, Customer Support, Legal, Accounting interested in

  • Expanding the dialogue with the global customer base
  • Accelerating growth and adoption in international markets
  • Developing products through global collaboration
  • Understanding how to leverage intellectual capital to support globalization
  • Increasing the flow of information and knowledge exchange across the global corporation

Mid-level Managers interested in

  • Reducing overall translation costs while increasing the volume of content that is translated
  • Improving translation productivity
  • Increasing the amount of content that available in translated form

Agenda

09:00-09:15 Welcome & Introductions

09:15-09:30 Translation automation in a market perspective

Changing market conditions and MT applications

09:30-10:00 Evaluating quality of MT – Different approaches

10:00-10:30 SMT in a historic perspective

Different approaches RbMT vs SMT

10:30-11:00 Coffee Break

11:00-11:45 Fundamentals of SMT

How it works, how it’s different, how high quality translations are generated and why data is important.

SMT and developing a continuous improvement cycle

11:45-12:15 Linguistics based data cleaning strategies to ensure continuous quality evolution

12:15-12:30 Evaluating MT

Hands-on exercise using industry-standard evaluation metrics participants will score the quality of machine translated texts and discuss how useful different examples of SMT output are.

12:30-13:30 Lunch break

13:30-14:00 User cases of SMT: MSFT & Others

14:00-15:00 Requirements for Success

  • What is clean data and how to optimize data for SMT training
  • Defining a quality evolution strategy and understanding what kinds of resources and tools are required.

15:00-15:30 Break

15:30-16:30 Discussion

  • Implementation scenarios and deployment models and the impact of crowd sourcing and the future of SMT
  • Integration with TMS systems
  • Integration with Content Management Systems
  • Selling the value of a large SMT project to management

16:30-17:00 Open Forum and Closure

About Kirti Vashee

Kirti Vashee is a seasoned sales and marketing executive of technology products who has built a reputation as an evangelist for SMT technology. He has been a prominent and accomplished speaker on automated translation technology. Kirti is known for his vision on the growing importance of automated translation technology in furthering international trade and communication. In addition to his worldwide partnership building experience, he has excellent knowledge of the knowledge management and translation industries. Previously, Kirti was Vice President of Sales & Marketing at Language Weaver, an SMT pioneer, where he was responsible for architecting their worldwide business development strategy and pioneering several SMT applications with Fortune 50 clients. He was educated in India and the US.

About Heidi Depraetere

Heidi has more than 20 years experience in the international localization and language technology industries. She is a founder of Cross Language, a privately owned consulting and systems integration company dedicated to translation automation technology. The company brings together a team of professionals with expertise in process engineering, computational linguistics, knowledge management, business automation and localization. Founded in 2002, Cross Language is now a leader in independent translation automation solutions and works for a growing number of global companies.

Heidi is familiar with the evolution of the industry from CAT tools to the newer globalization technologies, including workflow systems, machine translation and hybrid translation solutions. She has created and worked with international teams from different nationalities throughout Europe and the United States. Heidi is based in Ghent, Belgium and was educated in Belgium and Germany. She is fluent in English, Dutch, German and French.

About LISA Workshops

LISA Workshops make use of presenters with real-world experience in the subjects they teach about. Workshops do not pitch any products or services, but instead provide a general overview with candid and practical evaluations of tools and products that impact your business.