With

WITH can be used to project some expressions, and possibly compute some aggregations, from what your query has computed up to a point before starting the next part of your query. Although there are other common use cases two very common cases of using WITH is: (1) to compute a numeric result, often through an aggregation, that can be used as a predicate in the following parts; and (2) performing a “top-k” computation on a some query, before performing other querying on those top-k. We give examples of each.

We will use the example database for demonstration, whose schema and data import commands are given here.

Using WITH for aggregations

The following query returns all Users whose ages are greater than the average age of Users in the database. This can be done in two steps: (i) compute the average age of Users in the database and assign it to an alias avgAge; and (ii) use avgAge in a following query part in a comparison predicate. This is done in the following query:

MATCH (a:User)
WITH avg(a.age) as avgAge
MATCH (b:User)
WHERE b.age > avgAge
RETURN *;

Output:

---------------------------------------------------------
| b                                         | avgAge    |
---------------------------------------------------------
| (label:User, 0:1, {name:Karissa, age:40}) | 36.250000 |
---------------------------------------------------------
| (label:User, 0:2, {name:Zhang, age:50})   | 36.250000 |
---------------------------------------------------------

Using WITH for top-k computations

Suppose you want to return the Users that the oldest Users in your database follows. This can be done in two steps: (i) compute the oldest person “a”; and (ii) find the Users that “a” follows and return.
This is done in the below query:

MATCH (a:User)
WITH a
ORDER BY a.age DESC
LIMIT 1
MATCH (a)-[:Follows]->(b:User)
RETURN *;

Output:

-------------------------------------------------------------------------------------
| b                                       | a                                       |
-------------------------------------------------------------------------------------
| (label:User, 0:3, {name:Noura, age:25}) | (label:User, 0:2, {name:Zhang, age:50}) |
-------------------------------------------------------------------------------------

The part of the query until LIMIT computes the oldest user Zhang (aged 50) and then the last MATCH (a)-[:Follows]->(b:User) RETURN * returns the Users that Zhang follows (there is only one such User, who is Noura.)