Thursday,August 15,2019 | Press Release
15 August 2019, Shanghai UTH International announced the acquisition of Chengdu-based ePoolink for RMB 51.3 million (around USD 7.3 million). The latter, established in 2014, is a language tech providing cloud-based TMS, translation, interpreting, DTP, and data services, with a revenue exceeding RMB 5 million (USD 0.7 million) in 2018.
Committed to Parallel Language DataOriginally a language data company, UTH’s exclusive language database features 15 billion parallel translation segments across over220 languages, including most spoken languages like Arabic, Chinese, English, French, German, Spanish and Japanese, and less-spoken ones like Estonian, Romania, Slovenian, Lao, and Khmer. Boasting a full coverage of languages across the world, UTH has drawn significant interest from venture capitals, and has won cooperation from a considerable number of research institutions, translation agencies, government agencies, and large tech firms.
Exploring Solutions Powered by Quality DataDedicated to AI and data-driven solutions, UTH has continuously facilitated its language data business through R&D on a variety of products in MT, language teaching, and corpus researches, which in turn sharpens its insights in the field of big language data and artificial intelligence. A number of handy tools have already been put into use: 1) Sesame Lingofor language teaching and training; 2) Sesame Search, a corpus product featuring multiple dimensional data classification, a search function, intuitive data presentation and patented language processing technologies; 3) a series of domain-specific corpora for users in legal, finance, life sciences, and manufacturing fields. In mid-2018, UTH made several acquisitions to expand its business scope to e-learning, smart writing, and data mining. Partnered with Sogou, 5 mid-sized LSPs in 2017 and several more in 2018, UTH has established a unique business eco-system and has supposedly become the largest translation database in China, attracting an increasing number of leading AI, e-commerce, MT, and cross-border business firms reaching out for cooperation.
Efficient Mechanism in Data ProcessingUTH has embarked on a pioneering road similar to TAUS, yet is different from the latter in data collection and processing approaches. The TAUS Data Association (TDA)works through a data-sharing mechanism, where the control of its data quality is largely dependent on data-owners’ integrity and internal quality control. At UTH, data collection and processing is conducted by dozens of professional in-house corpus linguists and engineers, along with approximately 400 contracted linguists who work meticulously in language data collection, alignment and annotation under rigorous internal quality rules and supervised by trained testers. Thanks to this approach, UTH has formulated a comprehensive data quality management framework with reference to LQA models and ISO standards, which supports its data processing tools to help improve efficiency in its data processing workflow.
Domain-specific In-context Language Data for Optimum ResultsUTH’s focus on in-domain and in-context data is its major differentiator. The company’s language data resources are classified into three categories consisting of 15 Level I domains, 41 Level II domains and 178 Level III domains, to better fit the variety of demands of clients from both the language industry and the academic world. As the largest repository of parallel texts in China, UTH has been cooperating with LSPs, freelance translators, language service buyers, and several big data organizations, orchestrating high-quality data flow among these partners and turning the immobile language data into flowing values. As a hub in data exchange, filtering, and processing, UTH has become an indispensable part and a booster in this trade.
Riding the Wave of NMT: Towards a Brighter FutureWith the increasing application of NMT on Tweeter, Facebook, WeChat, QQ, and other UGC platforms, as well as the industrial application of MT in services that help people communicate with each other across language barriers, translation is becoming a crucial business energizer. Meanwhile, as the technological advantages of forerunners such as Google are diminishing, the gap in the translation quality among NMT vendors is narrowing. In this context, clean and fine-tuned translation data, which is of vital importance in improving NMT results, is playing an ever more crucial part in this fierce competition.
About UTH InternationalEstablished in 2012, UTH International is committed to providing innovative multilingual language solutions for customers across all industries.