Expected to Expand the Scope of Classical and Modern Literature Translation Through the System
Expansion of the project to develop automatic translation models for various classical texts

Systran International announced that it has successfully completed the 'Establishment of an AI-based Automatic Translation System for Classical Literature' project for the Korean Institute of Classical Translation, which is a task of the Ministry of Science and ICT (Korea Information Society Agency)'s '2017 ICT-based Public Service Promotion Project' that has been underway since last June.

Systran explained, “This AI-based automatic translation solution for classical texts, being introduced for the translation of classical and modern literature, is considered one of the examples of the convergence of humanities/social sciences with science and technology, and is expected to contribute to securing Korea’s future growth engines and academic assets.”

This translation system was completed through the training of more than 30 AI-based automatic translation models based on a total of 350,000 corpora from the Seungjeongwon Ilgi provided by the Institute for Classical Translation covering the reign of King Yeongjo from the year of his accession to the throne until March of the fourth year, as well as Systran's own corpus.



Neural network translation technology applied to automatic translation systems identifies unique patterns of correlation between the source text and the translation, and generates a translation model based on that data. This translation model autonomously learns information on the correspondence relationships between the source text and the translation, and generates the final translation by searching for and determining the most suitable equivalent word relationships during the actual translation process.

In addition, Systran has developed and integrated a tokenizer for classical literature into the system that separates classical Chinese characters into meaning units so that the AI translation engine can easily recognize classical literature. Furthermore, by 2018, Systran plans to add a corpus of classical literature and expand the machine learning length of translation texts so that long sentences from classical literature can be smoothly utilized in the initial translation process.

Systran plans to carry out continuous advancement projects, such as expanding the automatic translation model for classical literature targeting the original text and translation databases of the Annals of the Joseon Dynasty, Ilseongnok, and literary collections currently serviced in the Korean Classics Comprehensive DB, to assist classical translators with their initial translations and to implement an automatic translation service that can be used in the work and research processes of institutions and academic departments related to classical literature translation in the future.

Systran CEO Ji Lucas stated, “Systran is highlighting the company’s potential by successfully completing not only translation between languages but also the translation industry that connects the past and the present,” and explained, “As the world’s No. 1 intelligent language processing company, we will continue to strive for the advancement of various research fields, including historical research, and strengthen our business capabilities.”