Deploying Machine Translation in an Industrial Setting: Challenges - PowerPoint PPT Presentation

Deploying Machine Translation in an Industrial Setting: Challenges and Opportunities

Opportunities

Opportunity 1 On-line spending power — 2020 - $16 trillion — English only! ( Source: PracticalEcommerce ) Industry without borders — 2020 - $17.5 trillion — Non-English! ( Source: PracticalEcommerce )

Opportunity 2 73% of consumers more likely to purchase a product in their own language ( Source: EU Study of 2,430 web consumers in 8 EU Countries )

Opportunity 3 9 out of 10 internet users will visit websites in their own language when given a choice ( Source: EU Study of 2,430 web consumers in 8 EU Countries )

Opportunity 4 42% of consumers never buy a product/service in other languages ( Source: EU Study of 2,430 web consumers in 8 EU Countries )

Opportunity 5 The Comfort Factor Consumers prefer not to have to work too hard to get information that shapes their purchasing decisions ( Source: EU Study of 2,430 web consumers in 8 EU Countries )

Opportunity 6 40% of current Ecommerce Companies turn away international orders ( Source: QNET Market Report 2015 )

Lots and Lots of Opportunity — Selling Seamlessly Across Language Boundaries — 42% of consumers never buy a product in other languages — 40% of current eCommerce Companies turn away international orders — Offering Personalisation — 73% of consumers more likely to purchase a product in their own language — 9 out of 10 internet users will visit websites in their own language when given a choice — Building a Trustworthy eCommerce Brand Consumers prefer not to have to work too hard to get — information that shapes their purchasing decisions

An undeniable truth… 1 2 3 4 5 6 Multi-lingual Digital Age

Challenges

Challenges Some Interesting Challenges Building CMT Systems that Scale (massively)

Interesting Challenges Measuring Quality For Development Convergence of BLUE, TER, F-Measure mechanisms For Translators SDL Trust Score, KantanAnalytics

Interesting Challenges 1 3 Standards - MTQ Make it Predictive Establish a standard for Move beyond MT Quality comparative methods 2 Business Model Compliment current standard

Interesting Challenges Integration Post-Editing Making it fit into Making it fit into CAT workflows Environments Business Models Making it fit into business models

Interesting Challenges 5 Pretranlation is the future! No difference between MT and TM 4 6 CAT Tools are dead Intelligent Convergence Post-Editing is the Fully Integrated TM, MT future! (more than 1), Glossary etc.

Interesting Challenges Models Static vrs Fluid Phrase, Hierarchical, On-the-fly Neural etc. Which is best? improvements Processing Speed Every micro-second is sacred

Interesting Challenges 7 9 Super Fast Speed Adaptive MT The new must-have for all New modelling need MT systems to be superfast 8 Feedback system Translator needs to see improvements real-time to boost confidence in using MT Systems

Thank you…