(Contributors: Wang Pengjie, Dai Ningyan) On November 5, 2024, the Workshop on Multi-Modal AI-Assisted Deciphering of Oracle Bone Inscriptions was successfully held at Huazhong University of Science and Technology (HUST). Hosted by the School of Software Engineering, the event featured prominent experts in the fields of oracle bones and artificial intelligence, including Song Zhenhao, a member of the Chinese Academy of Social Sciences, Gong Changwei, President of the Chinese Society for Pre-Qin History, Shu Zhan, Head of Tencent's SSV Digital Culture Lab, Chief Architect Wang Chaoyang, and team member Wu Pian. Other attendees included Liu Yongge, Director of the Ministry of Education Key Laboratory for Oracle Bone Information Processing, researcher Qiao Yanqun, and special researcher Xu Zhaoxuan. School Party Secretary Wang Shixian, Dean Professor Bai Xiang, and other faculty members participated in the event, which was chaired by Professor Liu Yuliang.
In his opening remarks, Wang Shixian warmly welcomed the attending experts on behalf of the School. He introduced the School's academic and research strengths and emphasized its strong support for leveraging artificial intelligence and related disciplines to advance oracle bone research, thereby contributing to national cultural development and fostering cultural confidence.
Professor Liu Yuliang presented the team's latest progress in using AI to assist oracle bone decipherment. He proposed a multi-dimensional approach encompassing form, sound, meaning, and reasoning to decode oracle bones. He also highlighted the critical role of AI technology in text semantic analysis and character recognition, while discussing the practical needs for interdisciplinary collaboration with the experts.
Song Zhenhao underscored the importance of function words and phonetic-semantic compound characters in oracle bone studies. He suggested that categorization and semantic analysis could improve the efficiency of decipherment, particularly in the application of text content and literature retrieval. He noted the significance of precision in text analysis and thematic categorization during the decipherment process.
Gong Changwei encouraged engineering researchers to deepen their knowledge of oracle bone studies. He proposed optimizing decipherment by utilizing homophones and phonetic loan words. He also pointed out that while AI models have great potential, their application in oracle bone research should undergo expert validation to enhance effectiveness.
Liu Yongge highlighted the dependency of oracle bone research on phonetic-semantic compound characters and semantic analysis. He stressed the necessity of collaboration with humanities scholars in interdisciplinary studies and advocated for partnerships in data standardization and compilation of rubbings to enable more systematic progress.
Qiao Yanqun emphasized the importance of semantic associations and contextual segmentation accuracy. He suggested collecting usage examples from existing oracle bone materials to support the supplementation of fragmented text and the categorization of content, bringing innovative perspectives to text analysis.
Wang Chaoyang introduced Tencent's work on digital culture projects, particularly the development of an open platform for oracle bone studies. He expressed a desire to collaborate with more institutions to build a digital research platform for oracle bones, advancing the exploration of ancient civilizations.
Bai Xiang proposed strengthening collaboration between humanities and information science experts. He recommended joint funding applications to attract more young scholars, especially those specializing in AI, to engage in oracle bone research and expand the field's societal impact. He also suggested promoting oracle bone research through events and competitions to inspire younger generations to pursue studies in this area.
This workshop provided valuable insights into the integration of AI and oracle bone research, laying a solid foundation for future collaborations. The School of Software Engineering remains committed to advancing fundamental AI theories and applications, with a focus on interdisciplinary efforts in medicine, humanities, and other domains. Earlier this year, a paper titled "Deciphering Oracle Bone Language with Diffusion Models," supervised by Professor Bai Xiang, won the Best Paper Award at the Association for Computational Linguistics (ACL) conference. This marks only the third time in ACL's 62-year history that a Chinese research team has received this top honor. The study utilized image-based generative models to develop a conditional diffusion model optimized for oracle bone decipherment.
