TAUS User Conference 2009

Program | Agenda | Venue: Portland | Reports

Home > Agenda

TAUS User Conference

The agenda has been shaped by over 300 industry insiders from around the world who contributed to the TAUS Innovation Roadmap report. The event will show how the best in the industry are dealing with the main opportunities and challenges highlighted by the report.

The theme "Profit of Sharing" encapsulates a new spirit of innovation sweeping through the industry. The program is unique in its foresight, format and focus. Join us at the TAUS User Conference 2009 to take the pulse on the future of translation.

Tuesday, October 27
17.00-18.00	TAUS Data Association Gold Founding Members meeting
18.30-22.00	TAUS Executive dinner (Invitation only)

Wednesday, October 28
08.30 -12.00	TAUS Data Association Annual General Meeting
14.15	Conference Registration
15.00	Welcome and agenda overview Jaap van der Meer, TAUS
15.15	The Localization Business Innovation Roadmap Jaap van der Meer, TAUS
15.45	Why are we here? Aiman Copty, Oracle Corporation, and Robert Lattus, Intel Corporation Aiman and Robert, who are responsible for some of the largest localization budgets globally, will debate the translation business from the perspective of corporations that operate in every corner of the world.
16.15	Open Translation Platforms Open source, open to multiple vendors, open to multiple technologies and tools Pecha Kucha style presentations by acrolinx, GlobalSight, Moravia, Lionbridge, and SDL to update on significant developments from platforms presented in Edinburgh and showcase the best of a new crop.
17.15	Close
18.30	Welcome reception and networking dinner

{/moshide}

Thursday, October 29
08.55	Translation Automation and Language Data Sharing Session introduction by TAUS
	Translation Automation
09.00	Navigating the best route Pablo Vazquez, Cisco Systems Heidi Depraetere, Cross Language Reviewing the process and presenting a matrix to answer three questions perplexing so many in the industry: When is machine translation the right option? How do you select the best machine translation solution for you? How do you leverage machine translation to get the best quality in the most efficient way?
09.20	Moses and ROI before year-end Mariani Massimiliano, Autodesk Autodesk is aiming to rollout the open source machine translation engine, Moses, and reach ROI before year end. Mirko will provide insight on how after many training and evaluation iterations Autodesk got the right data mix as well as details on a two-day productivity lab involving twelve translators from three vendors. The Autodesk lab workbench was inspired by Philip Koehn’s Caitra project, collecting per-sentence timing data. Content and MT engines simulated a 2008 production scenario. All twelve translators saw their throughput increase significantly, with six at least doubling it.
	Translation Automation with Language Data Sharing
09.40	Breaking new ground (part 1) Chris Wendt, Microsoft Corporation A use case on reaching new heights of machine translation quality using industry-shared language data. The presentation will show the process, the criteria, the mechanisms, and the automatic and human evaluation results for each step in the process, enabling the audience to make deliberate choices about how to enhance the composition of training data for their statistical machine translation installation.
10.00	Breaking new ground (part 2) Ryan Martin, Intel Corporation A case study on advanced, in segment, leveraging using shared language data, as well as experiences from using TAUS search.
10.10	Breaking new ground (part 3) Ilia Kaufman, KCSL Ilia will present findings of KCSL's Total leveraging study, involving 15 translators to assess quality, cost and productivity across 5 scenarios.
10.20	Making industry-wide sharing easier Yan Yu, TAUS Data Association TAUS Data Association development roadmap. Yan will provide an update on improvements to make members lives easier, translators work simpler, and tools to stimulate innovation.
10.40	Refreshment break
	Before and after Machine Translation Two panel discussions
11.00	Normalization of Translation Memories Moderator: Karen Combe (PTC) Ryan Martin (Intel), Chris Wendt (Microsoft), William Wong (Language Weaver), Olga Beregovaya (PROMT) This panel will use specific examples and use cases to provide much needed guidance on data normalization to answer the following questions: What is the difference between cleaning and normalizing data? How do you balance the often competing needs for tag free translation memories to train machine translation engines against the use of metadata for accurate translation? What is the best way to deal with language-specific style guides, which may produce mismatches between source and target? Best Practices in Post Editing Machine Translation Moderator: James J. Kim (Intel) Ray Flournoy (Adobe), Fernando Blasi (AlphaCRC), Chris Wendt (Microsoft), Peter Smith (SDL), Melissa Biggs (Sun Microsystems) Demand for post-editing is set to go through the roof. The panel will share their expertise to answer the following questions: Why should you care about post editing machine translation? What do you need to do to get started? How do you measure quality? How do you get the best results?
12.40	Lunch and Collaboration Pavilions
14.00	Speed Collaboration Companies with Collaboration Pavilions and then every other participant, on a first-come-first-serve basis, are given the opportunity to sell their collaborative ideas, products, services in 3 minute back-to-back presentations, to all other participants
	Innovation Use Cases
14.30	Taking Moses to market Manuel Herranz, Pangeanic Public launch of PangeMatic v2 - Domain specific MT engines for automotive, industrial, engineering, professional & consumer electronics and software sectors. Pangeanic, a medium sized LSP, partnered with Valencia's Polytechnic University, customized Moses for clients in 2008 and then began using industry-shared language data from TAUS Data Association to provide MT for more clients earlier this year. Manuel will present a number of client case studies in addition to giving an overview of PangeMatic v2. This innovation use case is a prime example of the results of knowledge sharing through TAUS reports and forums. Moses in the cloud Achim Ruopp, Digital Silk Road The complexity of setting up and customizing machine translation systems, along with the cost of MT licenses are significant barriers to a wider adoption of machine translation. Achim will demonstrate how running the pre-configured open-source statistical MT system Moses on Amazon's EC2 Cloud-Infrastructure can break down some of these barriers. The training of a basic system will show the ease with which MT systems can be trained. Achim will also explore the best criteria for selecting training data to train domain-specific MT systems, improving the quality of MT output for specific translation projects significantly. New business models for a new market Marco Trombetti, Translated.net At the most recent TAUS event, Macro achieved the rare feat of surprising the world's the largest translation buyers with the amount of development work a small company can do. Marco will give practical examples of the benefits and costs of sharing language data as well as the costs of not sharing. Translated.net offers industry-shared web-based translation memories, integration of MT, community-translation (where, translators are important contributors to the shared TMs), and published APIs.
15.30	Afternoon break
16.00	Speed Collaboration Companies with Collaboration Pavilions and then every other participant, on a first-come-first-serve basis, are given the opportunity to sell their collaborative ideas, products, services in 3 minute back-to-back presentations, to all other participants
16.30	Breakout session
17.15	Close

Friday, October 30
08.55	Localization and Support Convergence Session introduction by TAUS
	Localization and Support Convergence
09.00	Dealing with the "Content Explosion" Greg Oxton, Consortium for Service Innovation To capitalize on 'the profit of sharing' requires a comprehensive and integrated view of content. Both internal knowledge management initiatives and the use of social media and mass collaboration through on-line communities have created a content explosion. The "content explosion" is a challenge the Consortium for Service Innovation has been exploring. We will discuss the concept and use of a content framework to help understand, organize and benefit from this seemly chaotic environment. The content framework is model to map content types and their attributes. Use of the framework can help organizations make informed decisions about what content is important, how best to publish information, content overlaps and gaps and localization/translation requirements.
09.40	Machine Translation Recipe for a Customer Support Knowledgebase The Intel Machine Translation Cookbook Translation automation "chefs" from Intel Corporation present their best recipe from Intel's Machine Translation Cookbook on how to prepare a delicious statistical machine translation system that can be integrated into a Web site. Their recipe describes how to deliver un-edited, fully automated useful translation for consumption by either external or internal customers; including the equipment, utensils, ingredients, preparation steps, "cooking" techniques and serving suggestions. The Intel chefs will also review other critical aspects involved with creating a gourmet MT recipe, including cultural differences in taste and acceptance, consumer feedback, how to measure quality, and various nuances of presentation. Participants are advised to come hungry for knowledge.
	Community Translation and Support
10.00	Community Driven Localization: What are the possibilities? Sean O'Driscoll, Ants Eye View Sean will look at both the trends driving adoption of user driven support as well as explore specific ideas for how organizations can leverage these trends to transform global support and localization.
10.40	Small crowd, big wisdom: How to leverage crowdsourcing in localization Pablo Vazquez, Cisco Systems At a time when many support organizations are looking to find ways of leveraging the crowd and community to help accomplish large and difficult tasks. Pablo will show how a small crowd was "self defined", seeded, and with the creation of a set of tools produced content that is helpful for crowd members as well as for localization efforts across Cisco. The presentation will focus on Cisco's Support Wiki and the NetPro community. Both offerings seek to create a space where customers and partners can interact freely to produce crowd knowledge that everyone can benefit from.
11.00	Morning break
	Innovation Use Cases
11.30	Crowdsourcing for Adobe Christine Duran, Adobe Willem Stoeller, Lingotek Chris and Willem will present a leading-edge example of community translation for Adobe using the Lingotek platform. The focus will be on lessons learnt and how community translation has helped Adobe fortify their relationship the community and non-US users and partners to further their competitive position. Seven Dinners from One Chicken: Getting the most out of translation memories Alex Yanishevsky, PROMT In this highly focused presentation, Alex will cover the entire translation memory life cycle to outline how translation memories can do double, triple or even quadruple duty. This includes validation, extracting customized domain-specific dictionaries, creating language models, advanced leveraging, and machine translation using domain customized engines as well as evaluation of translations.
12.10	Collective wisdom
12.30	The New Agenda Jaap van der Meer, TAUS
13.00	Lunch
14.00	Adjourn

TAUS - Enabling better translation

TAUS User Conference 2009

JOIN OUR MAILING LIST

OTHER TAUS SITES

TESTIMONIALS

TRANSLATION AUTOMATION TIMELINE

OUR APPROACH