BCAITS Upaya-Truth Suite v1.0 (BCAITS 权实研读系统)

A Dual-Engine AI Framework for Buddhist Canonical Studies & Zen Hermeneutics
佛典AI研读的基础设施:从“善巧方便”到“学术实相”的双轨系统

📖 Introduction / 简介

BCAITS Upaya-Truth Suite v1.0 is a scalable, open-source AI processing framework designed for Buddhist Digital Humanities. Unlike generic translation tools, this suite is engineered with a specific hermeneutic architecture inspired by the Buddhist principle of "Upaya" (Skillful Means) and "Truth" (Ultimate Reality).

It consists of two specialized applications that leverage Large Language Models (LLMs) with distinct inference parameters (Temperature) and prompt engineering strategies to serve different research needs: from the intuitive realization of Zen Koans to the rigorous philological analysis of Canonical texts.

BCAITS 权实研读系统 v1.0 是一个专为数字佛学人文学科设计的开源 AI 处理框架。与通用的翻译工具不同,本套件基于佛教判教思想中的**“权”(善巧方便)与“实”(真实义理)**构建了双轨架构。

该系统包含两个独立的应用程序,通过差异化的推理温度(Temperature)与提示工程策略,分别服务于不同的研读需求:从禅宗公案的直觉参悟,到经律论文献的严谨考据。

🛠️ System Architecture / 系统架构

1. BCAITS Upaya (The Skillful) | 权智·禅宗善巧系统

• Version: v14.0
• Logic: High Temperature (0.7) - Creative & Intuitive.
• Focus: Zen Koans (e.g., Records of the Lamp).
• Methodology: Integrates Cognitive Science (Embodied Cognition) and modern scientific metaphors (e.g., Quantum Mechanics) to translate ancient koans into accessible modern experiences.
• 定位: 高热度推理(灵动)。侧重于禅宗公案,结合认知科学与现代隐喻,旨在通过“善巧方便”引发读者的直觉性参悟,重在“意象”与“心流”。

2. BCAITS Truth (The Actual) | 实相·佛典考据系统

• Version: v1.1
• Logic: Low Temperature (0.3) - Rigorous & Consistent.
• Focus: Canonical Literature (Sutras, Vinaya, Abhidharma, Commentaries).
• Methodology: Emphasizes Philology, Structural Analysis (Pan-jiao/Ke-pan), and terminology standardization. It minimizes AI hallucination to provide a reliable basis for scholarly study.
• 定位: 低热度推理(严谨)。侧重于经律论义理,强调文献学考据、科判结构分析(章句切分)与名相释义,最大限度减少AI幻觉,服务于严肃的学术梳理。

🌍 Multilingual Matrix / 多语言矩阵

Both systems support automatic segmentation and generation of separate files and contrast study editions in 12 languages, forming a "6x2" symmetrical linguistic matrix:

两个系统均支持 12 种语言 的独立文件生成与对照研读版自动生成,形成了“6x2”的完美对称矩阵:

1. Chinese (中文)
2. English (英文)
3. Japanese (日本語)
4. Korean (한국어)
5. Tibetan (བོད་ཡིག)
6. Pāli (Romanized)
7. Spanish (Español)
8. Russian (Русский)
9. French (Français)
10. German (Deutsch)
11. Hindi (हिन्दी)
12. Norwegian (Norsk)

⚠️ Critical Technical Note / 关键技术说明

Regarding Text Segmentation (关于文本分段):

This software includes a built-in ZenTextProcessor class, which uses Regular Expressions (Regex) to automatically segment raw text.

• Current Optimization: The segmentation logic is currently optimized specifically for the text structure of "Wudeng Huiyuan" (X80n1565) (e.g., detecting headers like "Chan Master", "Volume", "Case").
• Limitation: If your input text does not share a similar structure to Wudeng Huiyuan, the automatic segmentation may fail or function sub-optimally.
• Usage: This program is provided as a framework/sample. Researchers are expected to modify the regex patterns in the source code or integrate their own AI segmentation models to adapt to different corpora.

关于分段算法的局限性:

本软件内置了 ZenTextProcessor 类,使用正则表达式对原始文本进行自动分段。

• 当前优化:目前的分段逻辑是专门针对**《五灯会元》(X80n1565)** 的文本结构(如“禅师”、“卷”、“则”等标题格式)进行优化的。
• 局限声明:如果您的输入文本结构与《五灯会元》差异较大,自动分段功能可能失效或效果不佳。
• 使用建议:本程序仅作为框架/样例提供。研究者需根据自身研究的文献类型,修改源代码中的正则规则,或接入自定义的AI分段模型以适配新文本。