从CSV导入关系至Neo4j：无法创建多条关系

Alright, let's break down why you're seeing only disconnected node pairs instead of the interconnected graph you expected, and fix this step by step.

Core Issues #

There are two key problems with your current Cypher query:

1. Unclean node names from CSV whitespace: Your CSV has spaces after commas (like name1, name2), so line[1] will include a leading space (e.g., name2 instead of name2). Neo4j treats these as distinct node names, creating separate nodes for name2 and name2—that's why your graph looks like isolated pairs.
2. Incorrect MERGE scope: Your original query merges the entire path (u)-[:connected_to]->(q) as a single unit. This works if you only want to create a relationship if the entire path doesn't exist, but it doesn't properly ensure shared nodes are reused (especially when whitespace messes up node names).

Correct Import Queries #

First, your constraint is already set up correctly—great job there:

CREATE CONSTRAINT ON (u:User) ASSERT u.name IS UNIQUE

Now use one of these queries based on your needs:

Option 1: Single Relationship Per Node Pair (Deduplicate)

Use this if you only want one connected_to relationship between any two nodes, even if your CSV has duplicate rows:

USING PERIODIC COMMIT
LOAD CSV FROM '${file}' AS line
WITH line LIMIT 50000
// Clean up whitespace to ensure consistent node names
WITH trim(line[0]) AS nameA, trim(line[1]) AS nameB
// First, merge each node individually to ensure they're reused
MERGE (u:User {name: nameA})
MERGE (q:User {name: nameB})
// Merge the relationship to avoid duplicates
MERGE (u)-[:connected_to]->(q)

Option 2: Preserve All Relationships (Including Duplicates)

Use this if your CSV has valid duplicate rows that you want to represent as multiple relationships between the same nodes:

USING PERIODIC COMMIT
LOAD CSV FROM '${file}' AS line
WITH line LIMIT 50000
WITH trim(line[0]) AS nameA, trim(line[1]) AS nameB
MERGE (u:User {name: nameA})
MERGE (q:User {name: nameB})
// Create a new relationship for every CSV row
CREATE (u)-[:connected_to]->(q)

Why This Works #

• The trim() function removes leading/trailing spaces from each CSV field, ensuring name2 and name2 are treated as the same node.
• Merging nodes individually guarantees that existing nodes are reused instead of creating new ones for every row.
• Separating node and relationship operations gives you control over whether you want deduplicated or multiple relationships.

Extra Tips #

• If your CSV uses a different delimiter (like tabs), add FIELDTERMINATOR '\t' to the LOAD CSV line.
• Adjust the PERIODIC COMMIT batch size (e.g., USING PERIODIC COMMIT 10000) if you're importing extremely large files to optimize performance.
• Test with a small subset first (reduce the LIMIT value) to verify nodes and relationships are created correctly before full import.

内容的提问来源于stack exchange，提问作者Traveling Tech Guy