Annotated catalog for the Kanseki Repository and beyond
The resource offered here is the result of instructing an AI agent to write cataloging notes for the texts. This agent was nick-named Bunkankun and this is used in the name of the site to remind users every time they visit that the information available here has to be read even more critical than usual, since such an agent can go off the rail at any time without warning. |Some Information about the making of the catalog. Ask Bunkankun is not a Chat interface, but rather has precooked answers to questions about the texts users might have, even if they don’t know it yet.
Contents
Overview of the Kanseki Repository Knowledgebase.
- KR1 經部 Jīng bù — Classics. Confucian classics, ritual, music, philology, and their commentarial traditions. 816 texts.
- KR2 史部 Shǐ bù — Histories. Dynastic history, chronology, geography, offices, statecraft, biography, bibliography, and historical criticism. 789 texts.
- KR3 子部 Zǐ bù — Masters. Philosophical masters, medicine, astronomy, mathematics, arts, mantic arts, encyclopedias, and miscellanies. 2485 texts.
- KR4 集部 Jí bù — Collected Works. Literary collections, anthologies, criticism, lyric forms, and narrative or extra-historical writings. 1812 texts.
- KR5 道藏 Dàozàng — Daoist Canon. Daoist scriptures, ritual manuals, alchemical works, revelations, hagiography, and canonical supplements. 1674 texts.
- KR6 佛藏 Fózàng — Buddhist Canon. Buddhist sūtras, vinaya, śāstras, school writings, histories, reference works, and extra-canonical texts. 4732 texts.
Total: 12308 texts in 6 divisions.
Coverage
The Kanseki Repository(herafter Kanripo) has been developed since 2012 and first published in March 2016. It organizes almost 10 000 items of premodern Chinese texts in six top level categories and makes the full text available for search and other research inquires, including download of all textual content. After publication of Kanripo, my attention shifted to a different project, that aimed at providing a more comprehensive set of interactions with the texts, including annotating, translating and other analytic and comparative activities. This is a new web version of the venerable Thesaurus Linguae Sericae(hereafter TLS), started by Christoph Harbsmeier, and since 2019 available at 漢學文典. This featured a much smaller, but thoroughly edited and punctuated set of roughly 1000 textual items. For a short while, new texts were added to both repositories, but very soon only to the TLS. Notable additions, like a corpus of medical text of around 1200 items, or about 300 texts on Chinese mathematics landed only here. Altogether, there are around 2000 texts that are missing from Kanripo but cataloged here, this is the ‘and beyond’ mentioned above. In addition to this thematic extension, the geographical scope also goes beyond China with the new additions: The Medical texts are presented as an integral tradition of East Asia, documenting the mutual borrowing and enhancing of practices and texts. And the Buddhist commentary tradition is also finally presented with the previous lacking Japanese commentaries newly added.
Context
The impetus for starting this effort was the desire to consolidate these two resources. This required better bibliographic information about the texts, necessary to place new texts into the existing classification. It was also desirable to attempt to better date the texts, so that it will be easier to appreciate textual relations and patterns of development. At the Bunkan, we have a yearly intensive course program teaching cataloging of premodern Chinese texts — 漢籍 Kanseki to attendants from institutions across Japan that have such holdings. These institutions may participate in the Kanseki Database (hereafter Kanseki DB), a union catalog of premodern Chinese texts in Japan.
Using these cataloging methods as a cue, and earlier efforts like The Taoist Canon1 as a model, I conceived a procedure that could be followed by a generative AI agent to produce catalog entries from earlier collections of metadata, other external references where needed, but primarily from any internal evidence the text might offer, especially in bibliographical notes, pre- and postfaces or other paratextual sections.
The agent has been instructed to surface any problems or discrepancies that it discovers — information that will be used to produce a new resource that will combine the two websites into a new interface. It will thus not read as a polished publication, but rather as a collection of research notes, warts and all, pointing to contradictions, gaps and other imperfections.
Collaboration
This catalog is very much tied to the collection it catalogs, which is by design — after all it was created to improve that collection. In addition to browsing the notes on the web, they can also be used locally on your own laptop or even smartphone or other mobile device. They have been created and edited to be compatible with a popular Markdown editor and note-taking application Obsidian as a downloadable ‘Vault’ — this is how Obsidian calls such a collection of notes. Instructions on how to go about this are available here. Not only offers this a better reading environment with more sophisticated searching, it also allows you to edit any notes, correct errors or simply make additions that are relevant to your own research question. On top of that, it is possible to contribute corrections back to the origin using a mechanism that is used by (open source) software developers all over the world and is increasingly used in the Digital Humanities and Open Science / Open Data movement: the distributed version control system git via GitHub. See the Readme file for further instructions on how to set this up.
Christian Wittern, May 2026.
Footnotes
-
Schipper, Kristofer, and Franciscus Verellen, eds. The Taoist Canon: A Historical Companion to the Daozang. University of Chicago Press, 2004. ↩