Question

我知道Import tool，但在我的情况下，我必须阅读一行并将其分解为节点和关系。使用带有索引的定期提交的load csv查询，并导入2Million行，花费超过12小时。有没有办法让我使用上面提到的工具而不必将csv预处理成节点和关系？

以下是我使用的示例查询

CREATE INDEX ON :Patient(mrno);
CREATE INDEX ON :Location(city);
CREATE INDEX ON :Department(id);

USING PERIODIC COMMIT 1000
LOAD CSV WITH HEADERS FROM "file:///home/geralt/Desktop/Temp_Admission.csv" AS line
WITH line,
(CASE  WHEN line.MRNo='' OR line.MRNo='null'  THEN "BLEH" ELSE line.MRNo END, "NA") AS mrn,
(CASE  WHEN line.ID_Admit='' OR line.ID_Admit='NULL'  THEN -1 ELSE line.ID_Admit END,0) AS ID_Admit,
(CASE  WHEN line.DeptCode_Admit='' OR line.DeptCode_Admit='NULL'  THEN -1 ELSE line.DeptCode_Admit END,0) AS DeptCode_Admit,
(CASE  WHEN line.City='' OR line.City='NULL'  THEN "BLEH" ELSE line.City END,"NA") AS city

MERGE (p:Person { mrn: mrn}) ON MATCH SET p.DOB=line.DateOfBirth,p.gender=line.GenderDescription,p.prefix=line.PrefixDescription ON CREATE SET p.DOB=line.DateOfBirth,p.gender=line.GenderDescription,p.prefix=line.PrefixDescription
CREATE (a:Admission{HospitalName:line.Hospital,id:toInt(ID_Admit),unitId:line.UnitID_Admit,IPDNo:line.IPDNO,DateOfAdmission:line.Date_Admit})
MERGE(d:Department{id:toInt(DeptCode_Admit)}) ON MATCH SET d.name=line.DeptName_Admit
MERGE(l:Location{city:city}) ON MATCH SET l.country=line.Country,l.state=line.State


merge  p-[:Admitted]->a 
MERGE a-[:Located]->l

Answer 1

执行多次运行（您甚至可以并行执行多个浏览器或neo4j-shell会话）应该非常简单。

删除ON MATCH SET
你拼错了 o
您缺少:Person(mrno), :Admission(id)
您的个案陈述已关闭
您在匹配集时使用ON CREATE SET
您可以通过在WITH上仅使用要导入的字段运行不同内容来进一步优化导入，请参阅部门

这是您的固定/完整/多次运行导入脚本：

CREATE INDEX ON :Patient(mrno);


CREATE INDEX ON :Location(city);
CREATE INDEX ON :Department(id);

// additional indexes / constraints

CREATE INDEX ON :Person(mrno);

CREATE CONSTRAINT ON (a:Admission) assert a.id is unique;

USING PERIODIC COMMIT 100000
explain
LOAD CSV WITH HEADERS FROM "file:///home/geralt/Desktop/Temp_Admission.csv" AS line
WITH line,
CASE  WHEN line.MRNo='' OR line.MRNo='null'  THEN "NA" ELSE line.MRNo END AS mrno

MERGE (p:Person { mrno: mrno}) 
  ON CREATE SET p.DOB=line.DateOfBirth,p.gender=line.GenderDescription,p.prefix=line.PrefixDescription;


USING PERIODIC COMMIT 10000
LOAD CSV WITH HEADERS FROM "file:///home/geralt/Desktop/Temp_Admission.csv" AS line
WITH line,
CASE  WHEN line.ID_Admit='' OR line.ID_Admit='NULL'  THEN -1 ELSE toInt(line.ID_Admit) END AS ID_Admit

CREATE (a:Admission{HospitalName:line.Hospital,id:ID_Admit,unitId:line.UnitID_Admit,IPDNo:line.IPDNO,DateOfAdmission:line.Date_Admit});

USING PERIODIC COMMIT 10000
LOAD CSV WITH HEADERS FROM "file:///home/geralt/Desktop/Temp_Admission.csv" AS line
WITH distinct line.DeptName_Admit AS DeptName_Admit,
CASE  WHEN line.DeptCode_Admit='' OR line.DeptCode_Admit='NULL'  THEN -1 ELSE toInt(line.DeptCode_Admit) END AS DeptCode_Admit

MERGE (d:Department{id:DeptCode_Admit}) 
  ON CREATE SET d.name=DeptName_Admit;


USING PERIODIC COMMIT 10000
LOAD CSV WITH HEADERS FROM "file:///home/geralt/Desktop/Temp_Admission.csv" AS line
WITH line,
CASE  WHEN line.City='' OR line.City='NULL'  THEN "NA" ELSE line.City END AS city

MERGE(l:Location{city:city}) 
  ON CREATE SET l.country=line.Country,l.state=line.State;


USING PERIODIC COMMIT 10000
LOAD CSV WITH HEADERS FROM "file:///home/geralt/Desktop/Temp_Admission.csv" AS line
WITH
CASE  WHEN line.MRNo='' OR line.MRNo='null'  THEN "NA" ELSE line.MRNo END AS mrno,
CASE  WHEN line.ID_Admit='' OR line.ID_Admit='NULL'  THEN -1 ELSE toInt(line.ID_Admit) END AS ID_Admit

MATCH (p:Person { mrno: mrno}) 
MATCH (a:Admission {id:ID_Admit})
MERGE (p)-[:Admitted]->(a);

USING PERIODIC COMMIT 10000
explain
LOAD CSV WITH HEADERS FROM "file:///home/geralt/Desktop/Temp_Admission.csv" AS line
WITH
CASE  WHEN line.ID_Admit='' OR line.ID_Admit='NULL'  THEN -1 ELSE toInt(line.ID_Admit) END AS ID_Admit,
CASE  WHEN line.City='' OR line.City='NULL'  THEN "NA" ELSE line.City END AS city

MATCH (a:Admission {id:ID_Admit})
MATCH (l:Location{city:city}) 
MERGE (a)-[:Located]->(l);

将巨大的csv文件导入neo4j

1 个答案: