Go global with this Winning Combination – Content strategy and Machine Translation

Content Strategy and Machine Translation The Winning Combination

Webinar Content

The Prize What can be achieved with MT

MT & The Content Strategy How content strategy can be

adapted for MT

Q&A

40 minutes

The Prize

What is KantanMT?

Statistical Custom MT Platform

Cloud-based Extremely customisable Highly scalable for high

volume Fully-automated translations High speed, high quality

translations

Active KantanMT Engines

9,392Training Words Uploaded

434,189,753,677Member Words Translated

7,161,269,925

KantanMT

Software Industry.

Software Industry.

ObjectivesBuild MT engines for on-line helpPilot for English => GermanScale to multiple target languagesFocus on Translator productivity

ChallengesComplexity of online help contentVery tight production schedules

Pilot & Other Languages < 4 months

Software Industry.

ResultsHighly Scalable: +4 languages in 6

monthsHigh Quality Translation outputTranslator Productivity Increase: 30-50%Reduced Project Costs: 27-40% savingsLower Post-Edit Rates after KantanMT

Engine Retraining

Case Study available for download.

Technical Manuals

Technical ManualsObjectives

Languages: English->DutchPOC to validate rapid ROIFocus on translation productivitySolution must scale to 50 manuals pa. Integrate to existing L10N workflow

ChallengesClient had already built in-house MTOutput of in-house system very poor

Technical ManualsResults

Seamless integration into existing L10N workflowTranslation productivity: + 25% Reduced Post-Editing effort using PEXReduced project turnaround timeSignificantly higher quality translations achievedProcess will scale to meet future requirements

Case Study available for download.

MT & Content Strategy

Strategic Decisions

High Level Business Needs - Create a ‘Global Model’

Structural Elements What content? Where is it needed? What are the differences?

Governance Elements How will it be managed?

Strategic Decisions – where?

= 3 billion people

What Languages?

Government of India – Hindi in the Devanagari script plus English

Indian constitution lists 22 recognised languages

2001 Census – 122 major languages and 1,599 other languages

Also recorded 30 languages spoken by more than 1m people

Education Levels?

Adult Literacy Rate, 15+ both sexes. Source: Worldbank

World Average

Philippines

India

ChinaIndonesia

Nigeria

In Thailand for instance?

Official government survey : Average amount of reading by respondents – 7 sentences per year.

Website experience - No use of ‘more info’ button No product descriptions Use of imagery and video

Think about how best to present the information

Structural Model and MT

What ContentWhat Locales and

Languages?

What Content Types?

Sharing, Differences,

Transcreation?

MT?

It’s the needs of the business that influence these decisions. Make the decisions and then review the process for MT fit.

Governance Model and MT

Centralised Federated Hybrid

• One MT system makes sense to generate economies of learning across management and users. (Unless that provider cannot service the locale/domain.)

• Cloud based solutions can be accessed from anywhere!• Leverage local language/domain expertise if required

Getting Tactical about Content for MT

Shorter sentences Remove needless words Be consistent Use glossaries Avoid idioms and jargon Use correct grammar Controlled Language Identify Named Entities Stick to Unicode Standard -

UTF8

Get serious about TMs

MT depends on Training data

Cleanse your TMs

Training data can help to:Improve precision – bi-lingualImprove fluency – mono-lingualMaintain consistency – glossaries

Library data and pre-built engines* can help to supplement data.

(*KantanLibrary™ & KantanFleet™)

How is Quality Determined

Automated scores Build Analytics Linguistic Quality Review

But ultimately: The customer decides!! Build in your own kpis

Thank You.