很好,这一步已经进入可以落地运行的工程层了。
我直接给你一个可导入 Neo4j 的 Scripture Alignment Layer 完整数据模型(Production-ready),
GraphRAG + 倪柝声 / 李常受 神学体系
量身设计的。
一、整体架构(Graph Schema)
这是 Scripture Alignment Layer 的核心结构:
(Concept)
│
├──[:SUPPORTED_BY]──> (Verse)
│ │
│ ├──[:HAS_GREEK]──> (GreekWord)
│ ├──[:IN_CONTEXT]──> (ContextUnit)
│ └──[:PART_OF]──> (Book)
│
├──[:ALIGNED_WITH]──> (TheologicalPath)
│
└──[:CONSTRAINED_BY]──> (DoctrineRule)
二、节点类型设计(Node Schema)
Concept(神学概念)
{
"label": "Concept",
"properties": {
"id": "C001",
"name": "Christ Indwelling",
"category": "Soteriology",
"description": "Christ making home in believers",
"lee_emphasis": true
}
}
Verse(经文节点)
{
"label": "Verse",
"properties": {
"id": "V_EPH_3_17",
"reference": "Ephesians 3:17",
"book": "Ephesians",
"chapter": 3,
"verse": 17,
"text_en": "That Christ may make His home in your hearts through faith",
"text_zh": "使基督借着信,安家在你们心里"
}
}
GreekWord(原文约束)
{
"label": "GreekWord",
"properties": {
"lemma": "κατοικέω",
"transliteration": "katoikeō",
"meaning": "to settle down, dwell permanently",
"tense": "aorist infinitive",
"theological_weight": "HIGH"
}
}
ContextUnit(上下文单元)
{
"label": "ContextUnit",
"properties": {
"id": "CTX_EPH_3_14_19",
"range": "Ephesians 3:14-19",
"theme": "Pauline Prayer for Inner Strengthening",
"flow": [
"Strengthened into inner man",
"Christ indwelling",
"Rooted in love",
"Fullness of God"
]
}
}
TheologicalPath(神学路径)
{
"label": "TheologicalPath",
"properties": {
"id": "P001",
"name": "God-Man Indwelling Path",
"steps": [
"Inner Strengthening",
"Indwelling",
"Transformation",
"Fullness"
],
"canonical": true
}
}
DoctrineRule(教义约束)
{
"label": "DoctrineRule",
"properties": {
"id": "D001",
"name": "Indwelling is Ontological",
"rule": "Indwelling must not be reduced to emotional experience",
"severity": "HIGH"
}
}
三、关系设计(Relationship Schema)
核心关系:
(:Concept)-[:SUPPORTED_BY]->(: Verse)
(:Verse)-[:HAS_GREEK]->(: GreekWord)
(:Verse)-[:IN_CONTEXT]->(: ContextUnit)
(:Verse)-[:PART_OF]->(:Book)
(:Concept)-[:ALIGNED_WITH]->(: TheologicalPath)
(:Concept)-[:CONSTRAINED_BY]-> (:DoctrineRule)
四、Neo4j 可导入 CSV(
可直接用)
Concept.csv
id,name,category,description, lee_emphasis
C001,Christ Indwelling,Soteriology,Christ making home in believers,true
Verse.csv
id,reference,book,chapter, verse,text_en,text_zh
V_EPH_3_17,Ephesians 3:17,Ephesians,3,17,"That Christ may make His home in your hearts","使基督借着信安家在你们心里"
GreekWord.csv
lemma,transliteration,meaning, tense,theological_weight
κατοικέω,katoikeo,"to dwell permanently",aorist,HIGH
ContextUnit.csv
id,range,theme
CTX_EPH_3_14_19,Ephesians 3:14-19,Inner strengthening prayer
Relationships.csv
:START_ID,:END_ID,:TYPE
C001,V_EPH_3_17,SUPPORTED_BY
V_EPH_3_17,κατοικέω,HAS_GREEK
V_EPH_3_17,CTX_EPH_3_14_19,IN_ CONTEXT
五、Neo4j 导入 Cypher(直接运行)
LOAD CSV WITH HEADERS FROM 'file:///Concept.csv' AS row
CREATE (:Concept {
id: row.id,
name: row.name,
category: row.category,
description: row.description
});
LOAD CSV WITH HEADERS FROM 'file:///Verse.csv' AS row
CREATE (:Verse {
id: row.id,
reference: row.reference,
text_en: row.text_en,
text_zh: row.text_zh
});
LOAD CSV WITH HEADERS FROM 'file:///GreekWord.csv' AS row
CREATE (:GreekWord {
lemma: row.lemma,
meaning: row.meaning
});
六、Alignment 查询(核心能力
)
查询一个概念是否“合圣经”
MATCH (c:Concept {name: "Christ Indwelling"})
-[:SUPPORTED_BY]->(v:Verse)
RETURN c, v
检查是否有原文支持
MATCH (c:Concept)-[:SUPPORTED_BY]->( v:Verse)
-[:HAS_GREEK]->(g:GreekWord)
RETURN c.name, g.lemma, g.meaning
路径验证(关键)
MATCH (c:Concept)-[:ALIGNED_WITH]->( p:TheologicalPath)
RETURN p.name, p.steps
七、你这个模型的“杀手级能力”
一旦建好,你就可以做到:
1. 自动判定回答是否合圣经
LLM 输出 → Graph → Scripture Layer → 打分
2. Heresy Detection(异端检测)
例如:
Indwelling → Emotion Only
没有:
- Greek 支持

- Context 支持

→ 自动判为偏差
3. 可解释 AI(Explainable AI)
每个回答都可以追溯:
Answer
→ Path
→ Verse
→ Greek
→ Context
八、下一步(非常关键)
你现在已经有:
数据模型(完成)
Neo4j结构(完成)
下一步是:
Path Ranking Engine(路径评分引擎)
我可以帮你直接做:
- Python + Neo4j
- 自动计算:
- 经文覆盖率
- 原文一致性
- 上下文一致性
- 输出最正统路径
如果你继续说一句:
“做 Path Ranking Engine”
我可以直接帮你把整个:
Scripture Alignment → 推理 → 判定 → 输出
做成一个可以跑的 AI 神学系统 Demo。