Funding Scheme

Innovation and Technology Support Programme

Project Title

Machine Translation of IPO Document

Project Team 
(HSMC Staff) 

Dr SIU Sai Cheong, School of Translation (PI)

Prof Francis CHIN, Department of Computing (Co-I)
Dr Haiqin YANG, Department of Computing (Co-I)

Other Collaborating Parties

Alpha Financial Press Limited

Project Period

1-6-2018 to 31-5-2019 (on-going)

Funding Amount (HKD)

1,400,000

Abstract

This project leverages on the facilities of HSMC’s HK$7M RGC-funded Deep Learning Centre and the expertise of its School of Translation to enhance the performance of Machine Translation (MT) from English to Chinese for HKEX documents, such as IPO documents and company annual reports. IPO documents (over 100 per year) and annual reports (for about 2000 listed companies) are usually prepared by lawyers and accountants in English and need to be translated to Chinese as required by HKEX. The translation task, usually handled by printing companies, is costly (about HK$0.5M per job) and labor-intensive.

Many IT-related companies have been investing heavily on Deep Learning technology which, based on the neural network architecture and large training data, has shown great success in many applications, in particular MT. Restricted domain MT should have better performance in terms of quality, cost and speed and is ripe for commercialization. Since HK is a bilingual city and as China becomes internationalized, there is an urgent and growing need for translation between English and Chinese. IPO documents will be a testbed for the idea of restricted domain translation and the prototype can be readily extended to other domains (legal and government) and regions in China.