2023-10-23
浏览次数：20The splendid Chinese civilization of thousands of years has left behind a vast amount of ancient literature and materials. These written records reflect the situation of society at that time in many fields such as politics, military, economy, technology, education, and culture, carrying rich historical information and cultural inheritance. Nowadays, however, most people find it difficult to read and understand ancient texts fluently. The use of advanced AI technology can enable ordinary people to understand and comprehend ancient texts, and also provide technical support for exploring and utilizing the rich knowledge contained in ancient texts.The Deep Learning and Visual Computing Laboratory led by Professor Lianwen Jin has built a large digital humanistic model for thorough understanding of ancient books - the Tonggu Grand Model based on large language model technology. The Tungu Grand model is based on the grand model that the laboratory has won the champion in the EvaHan2023 International Ancient Book Translation Competition. The model is obtained by combining the automatically generated dialogue template with the rich big data resources accumulated by Professor Lianwen Jin's team in the field of ancient books and through the training of large model instruction fine-tuning technology.Moreover, the model integrates multiple tasks in the form of natural dialogue, making it more convenient and effective for the public to understand the traditional Chinese culture, and the form is more friendly and natural, which is of great help to the spread and development of Chinese culture.
This team also developed an industry-advanced ancient book document analysis and recognition system(related technology has won the first place in the track of Cultural Inheritance - Chinese Multi-scene Recognition in the first Digital China Innovation Competition in 2019 and won the only best algorithm ability award in the final, and the champion of the first Great Bay Area International Algorithm Example Competition - Ancient Book Image Analysis and Recognition Competition in 2022).Users only need to provide a picture of an ancient book, and the system can automatically identify and locate all the texts in it, and sort the identified texts in the correct reading order.The system also integrates ancient book reading (automatic punctuation) and text translation independently developed by this team, which can automatically add punctuation marks to the recognized text and translate it into modern Chinese so that modern people can understand the ancient Chinese.
The algorithm of this system has been optimized carefully to deal with various challenges that may arise in the real world of ancient books, such as book bending, slant, high text density and low resolution, which are difficult for other systems to handle.Therefore, the system has excellent practicability and robustness, which provides strong support for promoting the digitalization of ancient books and helps to inherit and carry forward the excellent traditional Chinese culture.
In addition, the team has developed a Yi document analysis and recognition system, which is designed to meet the challenges of ancient Yi images, and it can automatically and accurately locate and identify the Yi text in the image (output is given in a custom code).The Yi coding used in this identification technology is based on the industry's first ancient Yi basic coding database jointly released by Shanghai University, Shanghai Hehe Information Technology Company and Professor Jin’s team earlier this year.
Classical Chinese is the carrier of traditional Chinese culture, and AI ancient book image recognition and classical Chinese translation technology can help people improve the understanding of ancient Chinese history and promote the inheritance of excellent traditional Chinese culture.In addition, AI classical Chinese translation technology can promote international communication and understanding, reduce the cultural gap so that foreign readers can also understand Chinese history and culture through translation, enhancing China's cultural influence in the world.On January 25, 2017, the General Offices of the CPC Central Committee and the State Council issued the “Opinions on Implementing the Project of Inheriting and Developing Fine Traditional Chinese Culture”, a document issued by the Chinese government to build a strong socialist culture, enhance the country's cultural soft power, and realize the Chinese dream of great national rejuvenation, which makes important guiding plans for the implemention of the project of inheriting and developing excellent traditional Chinese culture.In April 2022, the General Offices of the CPC Central Committee and the State Council issued the “Opinions on Promoting the Work of Ancient Books in the New Era”, pointing out that Doing well the work of ancient books, protecting, inheriting and developing the precious cultural heritage of the motherland is of great significance to maintaining the Chinese context, carrying forward the national spirit, enhancing the country's cultural soft power, and building a strong socialist culture.AI ancient book character recognition and classical Chinese translation technology is of great significance to promote the inheritance and development of ancient book culture, carry forward the national spirit, enhance the soft power of national culture, and promote the technical progress in the fields of data mining, knowledge discovery, intelligent development and utilization of Chinese ancient books and cultural relics.