Local start-up harnesses the power of artificial intelligence to tackle monumental task of financial translation
- Targeting a major pain point in financial publishing, DeepTranslate’s translation service is now used by major financial institutions in Hong Kong, including HSBC, Guotai Junan and Ernst & Young.
- Seasoned professionals and academia came together to bring research to successful commercialisation.
[Sponsored Article]
In 2017, Professor Francis Chin was leading a research project at the Hang Seng University of Hong Kong, where he sought to cooperate with local companies on artificial intelligence (AI) translation. He met Colin Hong, who has worked in the financial printing and translation industry for more than 20 years and was working for a listed company at the time. One year later, they decided time was ripe for the research to be further developed and commercialised. Together with Beth Chan, who has worked as an investment banker and barrister, they founded DeepTranslate.
With Hong Kong’s requirement for bilingual communications in the financial, legal and government sectors, the demand for Chinese-English translation is growing steadily at 7 per cent per year. A veteran in financial publishing, Hong understands all too well what a monumental task the production of documents such as disclosures and initial public offering (IPO) prospectuses can be, especially during peak season when huge amounts of translation need to be completed within a very short time.
“Free platforms such as Google Translate, which are available in many languages and aim to only help you understand a document, have quality score of 30 to 40 which is not enough to be used for things like prospectuses. On the other hand, DeepTranslate is all about accuracy and style. We focus on Chinese-English translation and can achieve up to 70 to 80 in quality score,” explained Hong.
DeepTranslate reduces the need for checking numbers. It also keeps the original formatting so less time is needed for tidying up. He said: “The system is not meant to completely replace manpower, so clients still have to do the checks and final touches manually. However, with a big part of the translation process streamlined, clients have reported increased productivity of 35 to 55 per cent, freeing up their time to take on more jobs.”
One of DeepTranslate’s selling points is its ability to learn and follow a previous style, since different companies have different corporate styles, in addition to linguistic differences in mainland China, Hong Kong and Taiwan. DeepTranslate is in the process of applying for patent for this unique feature.
