Facebook has developed a kind of AI (人工智能) translating between any pair of 100 languages without depending on first translating to English.
Facebook’s system was trained on a data set of 7.5 billion sentence pairs collected from the web across 100 languages, though not all the languages had an equal number of sentence pairs. “What I was really interested in was cutting out English as a middle man. Globally there are plenty of countries where they speak two languages that aren’t English,” says Angela Fan of Facebook AI, who led the work.
For some language pairs, the new system shows prominent improvements in translation quality. For example, translating from Spanish to Portuguese is extremely strong because Spanish is the second-most spoken first language worldwide, meaning the researchers can get a lot of data. Translation between English and Belarusian (白俄罗斯语) also is improved because the AI can translate Russian into English, which shares similarities with Belarusian.
While the system isn’t in use on the social network site, Facebook plans to put it to work soon to manage the 20 billion translations made every day when people click “Translate” on posts (帖子) written in more than 160 languages. “Future work will be done on other languages,” says Fan, “especially for languages where we don’t have a lot of data, like South-East Asian and African languages.”
The work “breaks away from the English-centric models and tries to build more different models,” says Shelia Castilho of the ADAPT Centre at Dublin City University, Ireland. “That’s pretty new and different. But it’s hard to know if the AI does a right translation work,” says Castilho. “The users may not know the other language, so they cannot judge the correctness of the translation.”
【小题1】What’s Angela Fan’s purpose of doing the research?| A．To collect data of all kinds of people. |
| B．To make learning different languages easier. |
| C．To achieve her dream of becoming a translator. |
| D．To avoid using English as a middle man in translation. |
| A．There is much data on Spanish. |
| B．Portuguese is similar to Spanish. |
| C．Researchers have a good knowledge of the two languages. |
| D．Spanish is spoken by the largest number of people. |
| A．The safety of AI translation. | B．The correctness of AI translation. |
| C．The way of using Facebook. | D．The number of posts on the website. |
| A．Facebook AI Can Talk with the Users |
| B．Facebook Wants to Buy a New System |
| C．Facebook AI Can Translate Any Pair of 100 Languages |
| D．Facebook Has Lost Many Users |
