Remove unnecessary relationships between nodes? - neo4j

Question

I tried to build a graph model using data, here is the cypher query
LOAD CSV WITH HEADERS FROM 'file:///y.csv' AS line
MERGE (a:Employee {empid:line.EmpID})
ON CREATE SET a.firstname = line.FirstName, a.lastname = line.LastName
MERGE (y:Year {year:toInteger(line.YearofJoining)})
ON CREATE SET y.month = line.MonthNamofJoining
MERGE (c:Location {city:line.City})
ON CREATE SET c.pincode = line.PinCode,c.county = line.County,c.state =
line.State,c.region = line.Region
MERGE (ag:Age {age:toInteger(line.AgeinYrs)})
MERGE (a)-[:AGE]->(ag)
MERGE (ag)-[:LOCALITY]->(c)
MERGE (c)-[:JOINING_YEAR]->(y)
I need to return all connecting path between four employees, so I tried below query
MATCH p = (a:Employee)-[:AGE]->(ag)-[:LOCALITY]-(c)-[:JOINING_YEAR]-(y)
WHERE a.empid IN ['840300','840967','346058','320954']
return p limit 25
But the result i got correct but there are many unnecessary paths. i am uploading the resulted graph please check and correct where i am doing wrong.resulted image

Answer 1

There are potentially several things to fix in the import query:
The year nodes are misleading. I think you should extract the month attribute to a separate node, like this:
MERGE (y:Year {year:toInteger(line.YearofJoining))
MERGE (m:Month {month:line.MonthNamofJoining})-[:MONTH_IN_YEAR]->(y)
Also, the modelling seems wrong. Currently, a Location is linked to year (or soon: month in year) via JOINING_YEAR. An age is linked to a location. This does seem to make sense.
You probably want an intermediate node to represent the fact that an employee has a joined a location (given Neo4j doesn't support relationships between more than 2 nodes).
LOAD CSV WITH HEADERS FROM 'file:///y.csv' AS line
MERGE (a:Employee {empid:line.EmpID})
ON CREATE SET a.firstname = line.FirstName, a.lastname = line.LastName
MERGE (ag:Age {age:toInteger(line.AgeinYrs)})
MERGE (a)-[:AGE]->(ag)
MERGE (y:Year {year:toInteger(line.YearofJoining))
MERGE (m:Month {month:line.MonthNamofJoining})-[:MONTH_IN_YEAR]->(y)
MERGE (c:Location {city:line.City})
ON CREATE SET c.pincode = line.PinCode,c.county = line.County,c.state =
line.State,c.region = line.Region
MERGE (j:Join {empid:line.EmpID}) // need a property to merge on
MERGE (a)-[:JOINED]->(j)
MERGE (j)-[:LOCALITY]->(c)
MERGE (j)-[:JOINING_MONTH]->(m)
Your read query becomes:
MATCH p = (:Location)<-[:LOCALITY]-(:Join)<-[:JOINED]-(a:Employee)-[:AGE]->(:Age)
WHERE a.empid IN ['840300','840967','346058','320954']
return p limit 25
Unrelated formatting note:
the recommended case for attribute is camelCase (e.g. empId instead of empid) and for relation types is SNAKE_CASE (e.g. JOINING_YEAR instead of JOININGYEAR).
By convention, relation types are verbs more often than not.