13 Tech giants compete in language translation market AI translation
AI translation competition heats up
The rise of artificial intelligence (AI) is casting a growing influence on the translation industry, with four-leading tech firms — including Naver and Google — vying to gain the upper hand in the AI-converged language conversion market.
Microsoft, lesser-known for a translation business, is also pushing to differentiate itself with its own intelligent translation platforms, such as Microsoft Translator and Skype Translator. In particular, Skype internet call service users can watch translated sentences on a monitor in real-time.
The service covers nine languages including English, Spanish, Mandarin and German. The company is set to add Korean to the service soon.
It remains to be seen whether the multinational software giant can diversify its revenue channels with its translation services here, due to fierce rivalry against the nation’s top portal operator, Naver, and Google. The two tech companies are hyping up their own neural machine translation (NMT) based translation service to raise their user base which will serve as a foundation to improve translation accuracy.
Naver’s NMT translation app Papago took the lead last August and Google also joined the race by launching its upgraded Google Translation service in November.
“Papago has reached some 60 percent accuracy, an eye-catching growth from about 30 percent of our conventional statistical machine translation (SMT) service,” a Naver spokesman said. One noteworthy difference between the NMT and SMT is that the former keeps improving its accuracy, as it piles up more datasets, he said, adding that the firm aims to make the service reach 80 percent accuracy by 2019.
“We are continuing to expand our presence by forming multiple partnerships with companies and government bodies such as GS Retail, Seoul Metropolitan Police Agency and Hyundai Department Store,” said the official.
Papago is available in six languages — including Korean, English and Chinese — focusing mainly on the Korean-English translation and vice versa.
Naver has also added the NMT service to its Whale web browser. Users can access translated data by clicking a Papago button on the browser.
The NMT services allow users to translate different languages in a more natural manner, as the AI system, serving as a brain, analyzes and translates the whole sentence. This is a stark difference from the SMT which is based on phrase-based translation system.
Big data is the key to facilitate the human-like learning processes, according to him, which is why the company puts top priority in gaining more users and datasets for the success of Papago.
Another Seoul-based software player, Hancom, also seeks to widen its revenue streams by launching its GenieTalk translation service. It is a combination of the NMT and the rule-based machine translation (RBMT) which the firm says is a “hybrid translation service,” generating more accurate outcomes than NMT-only services.
The company says GenieTalk has focused on improving accuracy for non-frequently-used expressions to differentiate its service from other tech titans.
Seven language translations — including English, Chinese, Spanish and Korean — are available for the service, according to Hancom. The service has been named the official translator for the PyeongChang Winter Olympics next year, with the firm seeking to raise its relatively-weak global profile through the upcoming sports event.
“The GenieTalk robot, designed for the special event, will generate satisfactory outcomes for foreign visitors, allowing them to feel like they are talking to human beings,” a Hancom official said.