STANDARDIZING THE LOCALIZATION PROCESS DAVE KIM LOCALIZATION PRODUCER

STANDARDIZING THE LOCALIZATION PROCESS DAVE KIM LOCALIZATION PRODUCER SONY ONLINE ENTERTAINMENT

OVERVIEW Sony Online Entertainment International Operations and Development (created in 2001) Current Active Localization Projects Ever. Quest II, Planet. Side, Pirates of the Burning Sea, Free Realms, The Agency, DC Universe Online, Legends of Norrath, Free Realms TCG, Match Attax, Station. com Annual Workload Number of Updates: 50~60 updates New English Word Count: ~ 2. 0 Million words annually Total Translated Word Count: ~ 8. 5 Million words annually Total Word Count in Translation Memory Database Total English Word Count: ~ 15 Million words Total Translated Word Count: ~ 60 Million words

CAPABILITY MATURITY MODEL INTEGRATION (CMMI) Optimizing Focus on process improvement Quantitatively Managed Process is measured and controlled Defined Process for the Organization, Proactive Managed Process for Projects, Often Reactive Initial Unpredictable, Poorly controlled, Reactive CMM WAS DEVELOPED BY SOFTWARE ENGINEERING INSTITUTE (SEI), CARNEGIE MELLON UNIVERSITY (CMU)

EVOLUTION OF LOCALIZATION STANDARDIZATION Localization without Standardization • No Translation Engine • No Persistent Unique IDs • No Standard Translation Data Exchange Format • Ad Hoc Localization with Standardization • Integrated Translation Engine • English Strings with Persistent and Unique IDs • Streamlined Translation Data Exchange Format • Modern Localization Workflow Localization with Optimized Standardization • Translation Engine Captures Lookup Calls • Translation Strings with Status • Contextual Data Exchanged • Work Flow Automation with Feedback

LOCALIZATION WITHOUT STANDARDIZATION No Translation Engine No Persistent Unique IDs • Problem: Not able to handle dynamic text • Result: Poor quality translation • Problem: Used English to look up translation • Result: Translation was lost when the English changed from “Go!” to “Go!!!” Ad Hoc Translation Data Exchange Format • Problem: Different every time, chaotic • Result: Lost files, data errors, user errors Ad Hoc Localization Work Flow • Problem: No Translation Database. • Result: Inefficiency. Delays.

LOCALIZATION WITHOUT STANDARDIZATION No Translation Engine • 6~12 man months of engineering clean up No Persistent Unique IDs • Excessive retranslation. $10, 000’s Ad Hoc Translation Data Exchange Format • 12 -18 man months of localization clean up Ad Hoc Localization • Labor intensive and Work Flow error-prone

INITIAL LOCALIZATION PROCESS New Content to Localize Publish Dev and Bugs QA and Bugs Eyeballing the Data Prepare and Send Batches Translation and Queries

MATURATION • The first step is managing the chaos. You can get quite good at each of these steps in the chain. • Eventually you reach the limits of chaos. • The next step is standardization. Ø Standardization of process. Ø Standardization of data. Ø Standardization of tools. • At SCE standardization is LAMS, at SOE it is C*STAR.

IMPLEMENTING STANDARDIZATION No Translation Engine Integrated Translation Engine No Persistent Unique IDs English Strings with Persistent and Unique IDs Ad Hoc Translation Data Exchange Format Standardized Translation Data Exchange Format Ad Hoc Localization Work Flow Modern Localization Work Flow

BENEFITS AFTER STANDARDIZATION Integrated Translation Engine English Strings with Persistent and Unique IDs • Improved quality with dynamic grammar • Ability to track changes in the English text Streamlined Translation Data Exchange Format • Simplified import and export process Modern Localization Work Flow • Centralized translation database and tools

BENEFITS AFTER STANDARDIZATION Outsourced Translation 60% workload New Game Assets Exact Match Translation (70% Disc) 25% workload Incremental Translation (50% Disc) 15% workload Publish (15% Disc)

OPTIMIZATION • Standardization for each project. Ø Ø Ø Standardize the data (content and context). XLIFF, Java Properties, XML schemas Standardize the tools. Standardize the processes. • The next step is to measure and optimize your standardization. Ø Ø Create metrics and feedback systems. Optimize based on those metrics and feedback. • The goal is continuous improvement, in all areas: Ø Ø Ø Cost reductions. Quality improvements. Greater speed and agility.

OPTIMIZING THE STANDARDIZATION Integrated Translation Engine Captures Lookup Calls English Strings with Persistent and Unique ID Translation Strings with Status Streamlined Translation Data Exchange Format Contextual Data Exchanged Modern Localization Work Flow Automation with Feedback

OPTIMIZED STANDARDIZATION PM Team Publish Translation/QA Master Tool Game Design (new content) CS/Community

TAKEAWAY • Standardization = Efficiency Ø 40% of translations internally from tools. Ø Saving 2+ man years of engineering per MMO. Ø Turn around 105 K words in 10 days, pushing daily with no new builds from engineering. • Optimization = Quality Ø Proper grammar. Ø Better translations from fuller context. Ø Faster, more productive QA.