AI and DH - The first presentation of CBDB colleagues (2024-2025)

The China Biographical Database project colleagues presented their projects on May 15–16, 2025, in Cambridge.

The presentations on May 15 were held in CGIS Knafel K354. There were two sessions on this day. The first session focused on social network analysis (SNA). The presentations in this session were:

How Writing Constructs Buddhist Historical Memory: A Reflection Centered on Data of Tang-Dynasty Monks slides
Xiong, Dian (CBDB visiting scholar from Henan University)
This presentation takes the monastic community of the Tang dynasty as a point of departure. Drawing on an analysis of over 2,800 data entries, it examines patterns of typological selection, gender exclusion, and shifts in narrative structure within the historiography of Chinese Buddhism. Using Mazu Daoyi and the "Hongzhou School" as a case, the talk explores how Chan Buddhism employed a linear, patriarch-centered genealogical narrative to replace the earlier, multi-dimensional "Ten-Category" classification system. This transformation played a crucial role in establishing sectarian discursive authority within Buddhist historical writing.

Geography and Membership of Ming Dynasty Literary Societies slides
Fu, Hongxiao (CBDB visiting scholar from Zhejiang University)
This study is the first to employ Large Language Models (LLMs) for data extraction, demonstrating that this approach achieves both high accuracy and efficiency. It significantly reduces the time required to structure data while expanding the potential data sources for the China Biographical Database (CBDB).
The Ming Dynasty witnessed the largest-scale literary societies in Chinese history. This research focuses on over a thousand individuals whose records align with CBDB data, analyzing their geographical distribution and their proportional representation within these societies. This allows us to explore the correlation between literary networks and local affiliations (xiangdang 乡党), as well as potential connections between the timing of society formations and dynastic shifts.
By constructing social network graphs, we reveal the associational preferences of literary figures, shedding light on how Ming elites leveraged these networks to cultivate cultural and political influence.

The topic of the second session is letters:

Class, Place and Text: Networks of Relations in Ming Material Correspondence slides
Han, Mingliang (CBDB visiting scholar from Zhejiang University)
A large corpus of extant Ming Dynasty letters offers a crucial window onto the relationship between text and the world. Visualization tools enable holistic comparative analysis of these letters, revealing a unique landscape at both cultural and documentary levels. Within this corpus, a prominent component consists of over 800 letters addressed to Fang Yongbin. Analysis of the spatial distribution of these correspondents and the frequency of their letters systematically reveals the extent of late Ming Huizhou merchants' activity within elite society, offering a social cross-section for understanding the close interactions between different social strata during this period. Beyond this specific corpus, another cluster of correspondence from within the Suzhou community exhibits characteristics of uniform and dense distribution. The layered and diverse communication network within Suzhou was primarily driven by three social dynamics: kinship networks, master-disciple relationships, and cultural activities. These two major correspondence corpuses represent distinct historical cross-sections for understanding the mid-to-late Ming period. Handling thousands of such extant documents offers new perspectives for reshaping historical understanding. While intellectually stimulating, this endeavor also presents a kind of historical documentary conundrum. By analyzing multiple documentary contexts—including the historical circumstances of production and collection, authentication issues, provenance, and archival practices—we attempt to reveal the factors contributing to the selection and survival of these specific historical cross-sections.

Core Figures of the Ming Dynasty: A Spatiotemporal Analysis of the “Minggong” Epistolary Network
Chen, Hongan (CBDB visiting scholar from Zhejiang University)
This presentation is based on Guochao Minggong Hanzao國朝名公翰藻, a letter anthology printed in 1587. It focuses on the core figures in Ming-dynasty literary, cultural, and intellectual circles—collectively referred to as the “minggong” (eminent officials)—up to that point in time. By organizing information on letter writers and recipients, the study constructs a preliminary social relationship and geographical distribution network among these elites. While the full identification of recipients is still ongoing, this research highlights the communications between minggong themselves, analyzing their spatial and temporal connections and their historical evolution. The study also explores the broader historical and social contexts that shaped these interactions. In addition, the presentation demonstrates how LLM technologies assist in identifying recipient identities.

Becoming Visible: Mapping the Social Worlds of Ming Artists
Zhuang, Yuning (CBDB visiting scholar from Southeast University)
This project draws on data from the Chinese Artists Dictionary, for which a standardized workflow was developed to process the complex layouts of modern printed publications. In combination with correspondence data from the CSA, it extracts a subset of artists who appear in correspondence records and analyzes how artists established reputations and cultural status. The analysis shows that central figures in the network were often politicians or literati rather than typical practicing artists, and that the spatial distribution of artists closely aligns with that of Ming jinshi. These findings raise questions about whether the criteria used in the Dictionary of Chinese Artists for defining “artist” may be overly inclusive. To further explore how aesthetic authority was shaped through social interaction, this paper takes Wen Zhengming as a case study and reconstructs his network of relationships and modes of exchange using multiple data sources, including the Futian ji, Elegant Debts, and CBDB. While the network structures generated from different sources vary considerably, each provides a distinct lens through which to approach the structure and dynamics of the Ming artistic community.

On May 16, the presentations were hosted in CGIS S050. There were also two sessions on this day. The first session was: How Technical Scholars Without Domain Knowledge Contribute to a Humanities Project:

Visualizing Complex Kinship Networks
Queenie Luo (CBDB research assistant and graduate student in the Department of East Asian Languages and Civilizations)
This presentation introduces a newly developed kinship network visualization tool designed specifically for the China Biographical Database (CBDB). The CBDB contains millions of biographical records, including detailed kinship data such as father-son relationships and extended familial ties. Despite the richness of the dataset, there has been a lack of visualization tools tailored to represent the complexity of these historical kinship structures. Existing family tree visualizations are often too simplistic to capture the multi-generational, multidirectional relationships found in the CBDB. This tool addresses these limitations by implementing a scalable, genealogy-focused visualization framework that enhances both the accessibility and readability of CBDB’s kinship data. The presentation will demonstrate how to use this tool and seek adivses from the general public to improve and expand the functions on this platform.

Constructing a Name Disambiguation Model for Ming Dynasty Officials Using Large Language Models
Liu, Zhenling (CBDB visiting scholar from Peking University)
This study focuses on the problem of name disambiguation in historical biographical records, aiming to construct a classification model capable of determining whether two individuals with the same name refer to the same person. The dataset is based on the Chronological Chart of Ming Dynasty Officials, where each individual is represented as a structured record. Wikipedia serves as an external reference for disambiguation, and large language models are employed to perform semantic matching between the structured records and corresponding Wikipedia entries. Within this framework, a set of labeled record pairs is constructed, incorporating multiple features such as textual similarity, temporal overlap of office-holding periods, and hierarchical rank alignment. A binary classification model is trained to produce similarity scores for each pair. The resulting model enables the disambiguation of name instances embedded within this feature space.

Findings from Siku Quanshu Data Mining slides
Ashley Zheng (CBDB research assistant and graduate student from Harvard College)
This presentation outlines my learning process in data mining through three stages. I began by recreating the processed Suzhou data, which helped me fact-check the first batch of Siku data and apply a similar categorization to the second batch. I will discuss the challenges posed by variations in indentation, style, and color across batches and how we developed individualized solutions. I introduced new tags, such as headings and subheadings, to improve the precision for content-oriented tagging. The presentation also evaluates trade-offs between precision and code reusability, taking a closer look at limitations in choosing to sort data by style in the second batch. Finally, I will compare the pros and cons of packages like docx and pypandoc and share pattern recognition tips that might be applicable to future projects.

The second session is Geographic Information System (GIS):

Data Mining and Research on Pre-modern Chinese Dramatists in the Era of Large Language Models slides
Chen, Hongyan (CBDB research assistant and graduate student from Hubei University)
This report centers on the data mining and analysis of Chinese dramatists from pre-modern China, investigating how digital technologies, propelled by the rapid advancement of Large Language Models (LLMs), enhance research on dramatist communities. Employing geographic visualization and statistical analysis, it systematically elucidates the spatial patterns and historical transformation of dramatists’ geographical distributions, while examining their interactions with regional socio-cultural structures. Building on a comprehensive analysis, the report uses the Jiangnan region as a case study to explore the spatiotemporal pathways and mechanisms underlying its emergence as a hub for dramatic creation.

Data Mining and Research on Chinese Translators Through the Ages
HU, Yiyi (CBDB research assistant and graduate student from Fudan University)
This research focuses on translators across Chinese dynasties, adopting an interdisciplinary digital humanities approach to explore their quantitative distribution, spatial patterns, and historical evolution. By integrating data mining, Geographic Information Systems (GIS), and visualization techniques, the research systematically examines the spatiotemporal dynamics of translators, identifying trends in regional clustering and migration across different historical periods. A multidimensional dataset is constructed, incorporating biographical attributes such as birth and death years, native places, and translated works, which enables an in-depth analysis of the socio-political, cultural, religious, and institutional factors shaping the geography of translation practices.
Building upon the general statistical overview, the study selects key historical periods—namely the Tang, Song, Yuan, Ming, Qing, Liao, and Western Xia dynasties—for focused case studies of representative translator communities. These cases investigate institutional mechanisms, intercultural mediation, and knowledge transmission within specific historical contexts. By bridging macro-level patterns and micro-level narratives, this research aims to offer new digital evidence and analytical frameworks for the spatial-temporal study of Chinese translation history, while also laying the groundwork for the construction of a comprehensive knowledge graph of Chinese translators.

Who Drove the Reform: A Study on the Composition and Regional Distribution of Late Qing Mufu Members slides
YAN, Liyuan (CBDB research assistant and graduate student from Fudan University)
My presentation focuses on the core question of late Qing reform: 'Who drove the reform?'. My research centers on the Mufu of three key figures: Zeng Guofan, Zhang Zhidong, and Li Hongzhang. Using methods of prosopography and visualization, I systematically examine the staff members' regional origins, social class, and their participation in Mufu activities, such as military affairs, political affairs, and especially the Yangwu (Westernization) activities. By combining quantitative analysis with intellectual history interpretation, I hope to provide a clearer picture of the composition and regional distribution of the reform forces in the late Qing, as well as their underlying ideological motivations. This offers a new, data-driven perspective for understanding late Qing transformation. In this presentation, besides introducing the basic data, I will first review the existing researches on topics like the late Qing Mufu, the dynamics of reform, and its key participants. Then, I will introduce my current progress.

Professor Peter Bol, Director of the China Biographical Database Project; Dr. Tang, Kwok-Leong , Managing Director of the Harvard Digital China Initiative; Yitian Li, Project Manager of the China Biographical Database Project; and the presenters all contributed to a rich and engaging discussion.

Amanda Stickels, Assistant Project Manager of the China Biographical Database; Kevin Lin, Senior Software Engineer at Harvard Arts & Humanities Research Computing; Lin, Wenzhou, Visiting Scholar in the Department of East Asian Languages and Civilizations; Yang, Yujuan, Visiting Scholar at Harvard University; and Cheng, Chao, CBDB Data Science Contributor, also participated. Wang, Hongsu, Senior Project Manager of the China Biographical Database Project, moderated all the presentations.