Loading CSV into a clean Neo4j database

To perform a bulk import of nodes and edges into a clean Neo4j database, make use of neo4j-admin.

Loading via neo4j-admin import is very fast and performant.

Before loading, ensure that nodes.csv and edges.csv have proper headers.

Make sure that each column in nodes.csv has an associated header that defines how that column should be interpreted by the loader.

For example,

:ID,name:string,category:LABEL
x:1,x1,named_thing
x:2,x2,named_thing

:ID - tells the loader that the column is supposed to be treated as node id property. (required)
name:string - tells the loader that the column is supposed to be treated as node name property. :string tells the loader to treat the column value as a string type. If no type is provided then the value is treated as a string, by default.
category:LABEL - tells the loader that the column is supposed to be treated as node category property. Additionally, :LABEL tells the loader to treat this column as the node label. (required)

Similary, any additional columns can be defined with the headers.

Make sure that each column in edges.csv has an associated header that defines how that column should be interpreted by the loader.

For example,

:START_ID,:TYPE,:END_ID,relation,provided_by
x:1,part_of,x:2,part_of,test

:START - tells the loader that the column refers to the start node of an edge. The value should correspond to a node id. (required)
:TYPE - tells the loader to treat this column as the relationship type of an edge. (required)
:END - tells the loader that the column refers to the end node of an edge. The value should correspond to a node id. (required)
relation - tells the loader to treat this column as relation property of an edge.
provided_by - tells the loader to treat this column as provided_by property of an edge.

Similary, any additional columns can be defined with the headers.

Once both nodes.csv and edges.csv are prepared with their proper headers, run the neo4j-admin import:

neo4j-admin import \
--database=knowledge-graph.db \
--id-type=string \
--array-delimiter=";" \
--nodes nodes.csv \
--relationships edges.csv

Provide feedback