AIを活用した文字分析の展望 - 古代記録物の中の判読不能部分を中心に -

초록

In this article, I first overview the impact that the rapid development of IT and AI has on various sectors of society and then analyse how AI technology can be applied to the decipherment and restoration of ancient written materials, especially wooden tablets. AI has attracted attention as a technology that opens up possibilities for generating new interpretations based on the reconstruction of characters and pattern analysis. Thus, various research projects have been proposed for scientific research programmes in Japan. Examples include research aiming to restore damaged documents using the image recognition and text generation functions of AI, and research projects applying deep learning methods to handwritten characters in order to analyse their characteristics. Therefore, the present study seeks to distinguish it from those mentioned above. More specifically, I point out the effectiveness of an indirect approach that not only achieves the direct recognition of undeciphered characters but also supplements them based on their relationship with already deciphered characters. In future research, I hope that the AI analysis will be advanced in combination with contemporary international sources.

키워드