Query to get all nodes given set of nodes - neo4j

Question

I have company metric data which I want to query
----------------------------------------------------
|Metrics | Year | Qtr | Department | Value|
----------------------------------------------------
|Revenue| 2017 | Q1 | Dep1 | 2000045|
|Revenue| 2017 | Q2 | Dep1 | 2000046|
|Revenue| 2017 | Q2 | Dep2 | 2000047|
|Revenue| 2017 | Q3 | Dep2 | 2000048|
|Revenue| 2017 | Q3 | Dep3 | 2000049|
|Sales | 2017 | Q1 | Dep1 | 2000041|
|Sales | 2017 | Q1 | Dep2 | 2000052|
|Sales | 2017 | Q2 | Dep1 | 2000053|
-----------------------------------------------------
Now I model the above data like this
Year, Qtr and Department as nodes like
(d:DIM {name:"2012","type":"year")
Value as nodes like
(v:VALUE {value:2000053})
and metrics as relationships like
(d:DIM {name:"2012","type":"year") - [:REVENUE]-> (v:VALUE {value:2000053})
So for each record there might be three relationships with VALUE node.
Now comes the query part:
Given dimension the query should get the values, like given Year 2017, Qtr q1 should return values corresponding to this filter if I add Dep 1 then it should further filter the result.
I tried some queries like
Match (d:DIM)-[:REVENUE]->(v:VALUE)
where d.name in ["2017","q1"]
Return d,v
But this query provides UNION of 2017 and q1 not the intersection I am looking for.
And further, I might do group by using the type attributes.

Answer 1

There are a couple ways you could do this. While I'd personally recommend using separate node labels for Year, Qtr, and Department (maybe Metrics too), I'll use your current model.
The piece you need for an intersection is the ALL() predicate, which operates on a list of values. In this case, we'll collect all matching :DIM nodes. To make the match efficient, we'll want to match to v nodes from the first item of the list, then ensure v is connected to the remaining items in the list (faster than filtering from all :VALUE nodes).
match (d:DIM)
where d.name in ["2017","q1"]
with collect(d) as dims
with head(dims) as head, tail(dims) as dims
match (head)-[:REVENUE]->(v)
where all(dim in dims where (dim)-[:REVENUE]->(v))
return dims, v
Alternately, if you have APOC Procedures installed, you can make use of an intersection function:
match (d:DIM)-[:REVENUE]->(v)
where d.name in ["2017","q1"]
with d, collect(v) as values
with collect(d) as dims, collect(values) as allValues // list of lists
with dims, reduce(inter = head(allValues), values in tail(allValues) | apoc.coll.intersection(inter, values)) as values
return dims, values
EDIT
I remembered another approach that is likely to be faster. Give this one a try:
with ["2017","q1"] as dimInput
with dimInput, size(dimInput) as dimCnt
match (d:DIM)
where d.name in dimInput
match (d)-[:REVENUE]->(v)
with v, dimCnt, count(distinct d) as cnt
where dimCnt = cnt
return v