joyboseroy/emptiness-graph
收藏资源简介:
Emptiness Graph 是一个类型化的哲学知识图谱,专注于佛教空性(sunyata)教义,涵盖上座部(Theravada)、般若波罗蜜多(Prajnaparamita)、中观(Madhyamaka)和瑜伽行派(Yogacara)等主要传统。该数据集不是聊天机器人数据集,而是一个结构化哲学图谱,编码了从巴利三藏的“无我”(anatta)到龙树(Nagarjuna)的中观、寂天(Shantideva)的辩证法、瑜伽行派的“三性”以及般若经等概念架构。核心见解是:空性并非可检索的事实,而是一种可遍历的关系结构,该图谱使这种结构显式化且可查询。数据集包含两个层次:第一层是手工构建的哲学图谱(包括概念节点、边和语料库清单),每个边都有哲学注释;第二层是自动生成的段落索引(包括文本段落和段落到概念的链接)。数据集共包含25个概念节点、38个手工标注的哲学关系、16个源文本元数据、1,126个文本段落和416个段落边。关系类型包括本体论、逻辑/辩证、教义和实践等17种。数据集旨在支持计算人文学科、知识图谱和文本检索等任务。
Emptiness Graph is a typed philosophical knowledge graph centered on the Buddhist doctrine of sunyata, covering major traditions including Theravada, Prajnaparamita, Madhyamaka, and Yogacara. This dataset is not a chatbot-oriented dataset, but a structured philosophical knowledge graph that encodes conceptual frameworks ranging from anatta (non-self) in the Pāli Canon, to Madhyamaka thought developed by Nagarjuna, the dialectics of Shantideva, the "three natures" doctrine of Yogacara, and Prajnaparamita sutras. The core insight of this dataset is that sunyata is not a retrievable factual entity, but a traversable relational structure, and this knowledge graph makes such structure explicit and queryable. The dataset comprises two tiers: the first tier is a manually constructed philosophical graph, which includes concept nodes, edges, and a corpus inventory, with each edge equipped with philosophical annotations; the second tier consists of automatically generated passage indexes, including text passages and links between passages and corresponding concepts. In total, the dataset contains 25 concept nodes, 38 manually annotated philosophical relations, 16 source text metadata entries, 1,126 text passages, and 416 passage edges. There are 17 relation categories, including ontological, logical/dialectical, doctrinal, and practical types. This dataset is intended to support tasks in computational humanities, knowledge graph research, and text retrieval.





