When did automatic translation technology begin to be studied?

Automatic translation technology has been studied since 1954.

Machine translation originated from 1933. The idea of machine translation was put forward by a French engineer, G.B. Arthuni, and obtained the patent of translation machine. 1954, Georgetown University in the United States, with the cooperation of IBM, completed the English-Russian machine translation experiment for the first time with IBM-70 1 computer, which opened the prelude of machine translation research.

1966, the Advisory Committee on Automatic Processing of Parallel Macros (ALPAC) established by Cheng Juesan of the American Academy of Sciences published a report entitled Language and Machine, which denied the feasibility of machine translation and entered a period of decline. From 65438 to 0976, the TAUM-METEO system jointly developed by the University of Montreal in Canada and the Translation Bureau of the Canadian Federal Government marked the overall recovery of machine translation.

1993, Brown of IBM put forward a statistical translation model based on word alignment, and the method based on corpus became popular. In 2003, Koehn of Edinburgh University put forward a phrase translation model, which clearly or significantly improved the effect of machine translation and promoted industrial application. In 2005, Chang Wei further put forward the hierarchical phrase model, and at the same time, the research on translation model based on grammar tree has made great progress.

In 20 13 and 14, researchers from Oxford University, Google University and Montreal University proposed end-to-end neural machine translation, which initiated a new era of deep learning translation. On 20 15, the attention mechanism was introduced into the University of Montreal, and the neural machine translation reached the practical stage. In 20 16, Google GNMT was released, NMT system was launched in Iflytek, and neural translation began to be widely used.