Neo4j：使用Cypher创建具有唯一节点的图形

我有一个发送者和接收者表：

Table A:

+------+--------+                                                          
|sender|receiver|
+------+--------+ 
|     A|       B|
|     B|       C|
|     C|       D|
|     D|       A|
|     E|       B|
|     A|       D|
|     E|       C|
+------+--------+  

我想创建一个从“发送方”到“接收方”的连接图。例如，

A-（rel）->B

发送方列中的值也可以出现在接收方列中。但是，我不希望节点A、B、C、D和E在我的图中重复。它们只能在最终图形中出现一次

我试着这样做：

LOAD CSV WITH HEADERS FROM 'file:///graph' AS row
CREATE (:sender_node { sender: row.sender})

MATCH (n:sender_node)
WITH n.sender AS sender, collect(n) AS nodes
WHERE size(nodes) > 1
FOREACH (n in tail(nodes) | DELETE n)

LOAD CSV WITH HEADERS FROM 'file:///graph' AS row
CREATE (:receiver_node { receiver: row.receiver})

MATCH (n:receiver_node)
WITH n.receiver AS receiver, collect(n) AS nodes
WHERE size(nodes) > 1
FOREACH (n in tail(nodes) | DELETE n)

LOAD CSV WITH HEADERS FROM 'file:///graph' AS row
MATCH (from_var:sender_node {sender: row.sender}),(to_var:receiver_node {receiver:row.receiver})
CREATE (from_var)-[:rel]->(to_var)
RETURN *

---

我基本上分别删除了发送方和接收方节点中的重复项。但由于节点A、B、C和D同时出现在“发送方”和“接收方”列中，因此它们在图中出现了两次。我想更正此部分。

电子邮件关系的开始节点隐式地是发件人。

电子邮件

关系的结束节点隐式地是接收者。因此，发送方和接收方节点不应该有不同的标签，这是多余的，并且使数据模型过于复杂

相反，您应该只为这两种节点使用

User

标签。这也将帮助您避免重复，尤其是当您使用子句而不是

例如：

LOAD CSV WITH HEADERS FROM 'file:///graph' AS row
MERGE (sender:User {id: row.sender})
MERGE (receiver:User {id: row.receiver})
MERGE (sender)-[:EMAILS]->(receiver)
RETURN *