深度學習方法在中國佛教經典目錄分類中的應用

Application of Deep-learning Methods in the Classification of Chinese Buddhist Canonical Catalogs

本研究採用深度學習方法，自動對中國佛教經典目錄進行分類。這項研究旨在解決如何將新增的佛教文獻納入傳統分類架構，以及手動分類耗時且難以達成共識等問題。我們透過將CBETA電子佛典集成的新增文獻整合到傳統目錄結構中，提高了分類準確性，並深入探討自動分類錯誤的原因。同時，我們還發現不同經典類別之間的歷史聯繫，未來可用於調整和優化目錄結構，以滿足現代研究需求。本研究成果已獲CBETA研究小組採納，成為未來新文獻編目的參考工具，以確保佛典目錄系統與學術領域變化同步並保持實用性。

This research focuses on the classification of Chinese Buddhist scripture catalogs, employing advanced deep learning methods to develop an automatic classification mechanism. Chinese Buddhist scripture catalogs serve as vital tools for organizing and retrieving Buddhist literature, facilitating research in the field. However, with the continuous addition of new texts and the need to adapt to modern academic research requirements, traditional manual classification methods have become time-consuming and less effective.
In this study, we aim to address this challenge by harnessing the power of deep learning techniques. Our research not only involves the integration of new literature into existing catalog structures but also explores the reasons behind misclassifications in these catalogs. Additionally, we examine the inherent connections between different categories of scriptures to provide a comprehensive understanding of the catalog structure. Our contributions in this research encompass:
1. Automatic Classification: We pioneer the use of deep learning methods for the classification of Buddhist scripture catalogs, allowing for faster and more accurate categorization of texts. This automated approach significantly enhances the efficiency of catalog management.
2. Error Analysis: We delve into the reasons behind misclassifications in the catalogs, shedding light on common pitfalls and misconceptions in traditional classification methods.
3. Interconnections: We uncover the original interrelationships between different categories of scriptures, offering valuable insights for adjusting and optimizing the catalog structure to align with contemporary research needs.
4. Practical Application: The research findings are adopted by the CBETA research group as a reference tool for the cataloging of new literature in future editions of the Tripitaka, ensuring that the catalog remains up-to-date and relevant.
This research is a significant step towards modernizing the management of Chinese Buddhist scripture catalogs, making them more efficient, accurate, and adaptable to the evolving landscape of Buddhist literature.