FOREIGN LANGUAGE IDENTIFICATION & SEMI-MANUAL MACHINE TRANSLATION
About our Client: The client is a premier provider of contract analysis solutions and discovery management services. The company offers discovery consulting, data acquisition, secure hosting, analytics and predictive coding, and productions. Their proprietary technology, combined with PM expertise, provides defensible solutions for navigating complex data. To solicit their business needs, Kensium BPO was engaged for language identification and translation services on a vast data set.
By the numbers…..
- ~110,000 pages manually reviewed and translated
- 5 dedicated Foreign Language Translation specialists over a 4 week span.
- 11 common languages translated using Kensium’s proprietary software and 31 secondary languages processed through semi-manual translation.
Challenge: In June 2019, a long standing litigation support client contracted Kensium BPO with a unique requirement to perform language identification and machine translation on 110,000 pages of documents produced by Opposing Counsel in preparation for depositions. As a prestigious Am Law 200 law firm, the client had stringent security requirements that had to be addressed before the project could get underway.
Manual Language Identification is a tedious process requiring a thorough page by page review to detect multiple languages as well as employment of recognition software. Kensium professionals take individual snippets of text and enter them into the program to determine the correct language. As many languages of the same geographical region can have similar characteristics with only marginal differences to the common reader, painstaking care must be taken by experienced hands to ensure an accurate end product. After Identification, the files must then be segregated and processed either through machine translation for common languages or through a semi-manual process for more obscure dialects. In the latter process, portions of text are again manually run through the translator with the results delivered to a text file for client review.
Solution: Working directly with firm’s Information Security officers, Kensium’s IT staff underwent an exhaustive security assessment to assure 100% compliance. Upon completing the assessment, Kensium’s score of 98% (832.5 out of a possible 850) far exceeded the threshold required and the project was immediately started. In parallel with the security audit, an experienced team was deployed and a custom workflow designed that would push the results of each step into individual buckets based on the language processing required. Final translation text was assembled and quality checked thoroughly prior to delivery.
Results: Over a period of 4 weeks, the project team reviewed 110,000 pages, identifying 42 individual languages to be translated. Roughly half of the population was earmarked for machine translation with the balance processed by semi-manual means. Rolling deliveries were provided throughout allowing the project team to complete their review. These results were provided at a fraction of the time and/or cost of traditional translation services or internal resources. All work was done in an ISO 27001, GDPR compliant facility which exceeded the security standards set by the Client and has led them to call on Kensium repeatedly since.