Connecting one node to multiple

[jhayes8552]

Hello,

My problem is that whenever I try to import data via a large csv and connect to a central node, it seems to make many copies of the central node (see photo). The brown nodes are what I want connected to just one "Earth Justice" node. I realize I can merge duplicate nodes, but I would like to have it right as I load in.

My code is as follows:

:auto Using periodic commit
LOAD CSV WITH HEADERS FROM 'file:///EJ_Sample.csv' as row
CREATE (l:labid) SET l+=row
CREATE (o:Origin {name:"Earth Justice"})
MERGE (l)<-[:Created]-(o)

Thanks in advance.

[Joel]

MERGE can be a bit confusing to use, I agree, there are nuances that even the experienced run into again (and again). At a glance I think maybe you are explicitly creating the nodes (and creating duplicates after the first time)? Perhaps the MERGE statement is ok.

Reference:

Neo4j Graph Database Platform

Understanding how MERGE works - Knowledge Base

What is MERGE, and how does it work? The MERGE clause ensures that a pattern exists in the graph. Either the entire pattern already exists, or the entire pattern needs…

[ameyasoft]

jhayes8552:

LOAD CSV WITH HEADERS FROM 'file:///EJ_Sample.csv' as row
CREATE (l:labid) SET l+=row
CREATE (o:Origin {name:"Earth Justice"})
MERGE (l)<-[:Created]-(o)

Try this:
LOAD CSV WITH HEADERS FROM 'file:///EJ_Sample.csv' as row

MATCH (o:Origin {name:"Earth Justice"})
with o
CREATE (l:labid) SET l+=row
MERGE (l)<-[:Created]-(o)

[jhayes8552]

After copying and pasting your example, I get this error message.

Variable `row` not defined (line 5, column 25 (offset: 130))
"CREATE (l:labid) SET l+=row"

Doesn't quite make sense with me why that doesn't work.

[Benoit_d · Accepted solution]

Hi,

in order to insure that MERGE is recognizing the node you are addressing, you should instore a uniqness constraint on one property of the node-label before loading the data.

CREATE CONSTRAINT ON ( orig:Origin)  ASSERT org.name IS UNIQUE

then you will be able to MERGE this as "Origin" labelled node without recreating it, which means the first appearance of "Earth Justice" will create the node, all other will merge.

:auto Using periodic commit
LOAD CSV WITH HEADERS FROM 'file:///EJ_Sample.csv' as row
CREATE (l:labid) SET l+=row
MERGE (o:Origin {name:"Earth Justice"})
MERGE (l)<-[:Created]-(o)

Pay attention, that this means the node (o:Origin {name:"Earth Justice"}) should come from the file, which is not the case in your cypher: no reference to any column.
If the origin node allready exists, just make a match:

:auto Using periodic commit
LOAD CSV WITH HEADERS FROM 'file:///EJ_Sample.csv' as row
CREATE (l:labid) SET l+=row
MATCH (o:Origin {name:"Earth Justice"})
MERGE (o)-[:Created]->(l)

In some case, for instance if you are delivering a new origin, to which every node of the source have to be connected, but the name of this node already exist, you will have to

• create a temporary label e.g. tempOrigin,
• create a constraint on this label,
• load the data
• destroy the constraint (change "Create Constraint on ..." into "Drop Constraint on ...")
• relabel all nodes with label tempOrigin (should be only one) to label Origin
• delabel all nodes tempOrigin

A piece of cake 😉

[jhayes8552]

Thanks, appreciate the help. The Constraints was a good tip.

[jhayes8552]

One more thing: For the 3rd block of code, the error message I get is that

WITH is required between SET and MATCH 

The "With" doesnt execute the code, however. Thanks again for the help!

[Benoit_d]

move the SET to the end:

:auto Using periodic commit
LOAD CSV WITH HEADERS FROM 'file:///EJ_Sample.csv' as row
CREATE (l:labid) 
MATCH (o:Origin {name:"Earth Justice"})
MERGE (o)-[:Created]->(l)
SET l+=row

or make a "matched merge":

:auto Using periodic commit
LOAD CSV WITH HEADERS FROM 'file:///EJ_Sample.csv' as row
CREATE (l:labid) 
MERGE (o:Origin {name:"Earth Justice"})-[:Created]->(l)
SET l+=row