Abstract: The reuse of earlier recipes has been a contentious topic in both Daoism and medical history. Previous attempts at intertextual associations in Zhengtong Daozang 《正统道藏》, a collection of over 1500 Daoism literature, have been limited by traditional historical methods, restricting the range of literature used and the efficiency of the work, sometimes compromising the accuracy of comparisons. In this research, we propose a novel approach by incorporating new machine learning and computational graphing methods, such as n-gram and cosine similarity. As a pilot study, we focus on analyzing a Daoism recipe collection from the Wei and Jin Dynasty (c. 4 - 6th century CE), Taishang Lingbao Wufu Xu 《太上灵宝五符序》, by comparing it with the ctext based Hanfen 涵芬楼 version of Zhengtong Daozang 《正统道藏》. Through this analysis, we aim to expand the scope of comparison and gain a broader view of the formulation history of Daoism recipes, enabling us to address historical questions semi-quantitatively. Furthermore, we anticipate that this research will illuminate new pathways for interdisciplinary studies and foster future collaborations between scientific and humanities disciplines. However, we acknowledge the challenges we currently face, including the establishment of a machine learning-based workflow within a limited timeframe. The expected results of this study encompass three sections: (1) similar text detection, (2) quantified similarity assessment, and (3) visualization of the findings.

Team members

• Xiaoyi HE, upcoming master's student in Computer Science, Columbia University
• Yuchen FENG, Graduate student in History, Nanyang Technological University
• Chunnan LEE, PhD student in History, Huafan University
• Biyang WANG, PhD student in Archaeological Science, the Cyprus Institute

Supervisors

• Joe Dennis
• Donald Sturgeon
• Calvin Yeh