Neo4j 将独立查询与

我有两个单独的操作：

MATCH ()<-[r:waitsFor]-() 
  SET r.active = true 

---

WITH子句用于将返回的信息“结转”到查询的另一部分。在Neo4j中使用WITH null作为none并向前执行后续匹配容易受到笛卡尔乘积创建的影响。根据节点的数量，笛卡尔积爆炸可能是巨大的，需要花费大量时间

因此，如果第一个查询中的“r”在第二个查询中是缩小匹配范围所必需的，那么您应该使用WITH r。由于情况似乎并非如此（因为您将null用作none），因此这可能不是解决方案

---

在我看来，您的案例似乎在按顺序执行两个独立的查询。你应该看看这个条款。它用于连接两个查询并合并它们的结果。您拥有的选项是纯旧的

UNION

或

UNION ALL

（删除重复项）。

不要这样做。正如您所说，它们是独立的查询，所以请分别运行它们。按照现在运行它们的方式，第二个查询将对第一个查询中的每个匹配结果执行一次。我现在正在处理的一个小型本体图中的一个示例，有146个节点标记为

：OwlClass

。考虑下面两个查询

-查询1-

MATCH (c:OwlClass) 
RETURN COUNT(*)

MATCH (c:OwlClass) 
WITH null as n 
MATCH (d:OwlClass) 
RETURN COUNT(*)

MATCH (c:OwlClass)
WITH COUNT(c) as count_c
MATCH (d:OwlClass)
RETURN count_c, COUNT(d) as count_d

结果:

COUNT(*)
146
Returned 1 row in 102 ms

COUNT(*)
21316
Returned 1 row in 734 ms

count_c   count_d
146       146
Returned 1 row in 153 ms

执行计划：

==> ColumnFilter(symKeys=["  INTERNAL_AGGREGATE01b2d10b-7bf4-43b4-a67d-fbe826c7a4dd"], returnItemNames=["COUNT(*)"], _rows=1, _db_hits=0)
==> EagerAggregation(keys=[], aggregates=["(  INTERNAL_AGGREGATE01b2d10b-7bf4-43b4-a67d-fbe826c7a4dd,CountStar())"], _rows=1, _db_hits=0)
==>   NodeByLabel(label="OwlClass", identifier="c", _rows=146, _db_hits=0)

==> ColumnFilter(symKeys=["  INTERNAL_AGGREGATEe33d15c7-e357-4e32-9f7f-3fbc00dd58f6"], returnItemNames=["COUNT(*)"], _rows=1, _db_hits=0)
==> EagerAggregation(keys=[], aggregates=["(  INTERNAL_AGGREGATEe33d15c7-e357-4e32-9f7f-3fbc00dd58f6,CountStar())"], _rows=1, _db_hits=0)
==>   NodeByLabel(label="OwlClass", identifier="d", _rows=21316, _db_hits=0)
==>     ColumnFilter(symKeys=["c", "n"], returnItemNames=["n"], _rows=146, _db_hits=0)
==>       Extract(symKeys=["c"], exprKeys=["n"], _rows=146, _db_hits=0)
==>         NodeByLabel(label="OwlClass", identifier="c", _rows=146, _db_hits=0)

==> ColumnFilter(symKeys=["count_c", "  INTERNAL_AGGREGATE17be6af7-abbb-4ef7-8250-ff050d6ba7c6"], returnItemNames=["count_c", "count_d"], _rows=1, _db_hits=0)
==> EagerAggregation(keys=["count_c"], aggregates=["(  INTERNAL_AGGREGATE17be6af7-abbb-4ef7-8250-ff050d6ba7c6,Count(d))"], _rows=1, _db_hits=0)
==>   NodeByLabel(label="OwlClass", identifier="d", _rows=146, _db_hits=0)
==>     ColumnFilter(symKeys=["  INTERNAL_AGGREGATE67abdc54-9c44-4353-bf82-e4aff895141e"], returnItemNames=["count_c"], _rows=1, _db_hits=0)
==>       EagerAggregation(keys=[], aggregates=["(  INTERNAL_AGGREGATE67abdc54-9c44-4353-bf82-e4aff895141e,Count(c))"], _rows=1, _db_hits=0)
==>         NodeByLabel(label="OwlClass", identifier="c", _rows=146, _db_hits=0)

-查询2-

MATCH (c:OwlClass) 
RETURN COUNT(*)

MATCH (c:OwlClass) 
WITH null as n 
MATCH (d:OwlClass) 
RETURN COUNT(*)

MATCH (c:OwlClass)
WITH COUNT(c) as count_c
MATCH (d:OwlClass)
RETURN count_c, COUNT(d) as count_d

结果:

COUNT(*)
146
Returned 1 row in 102 ms

COUNT(*)
21316
Returned 1 row in 734 ms

count_c   count_d
146       146
Returned 1 row in 153 ms

执行计划：

==> ColumnFilter(symKeys=["  INTERNAL_AGGREGATE01b2d10b-7bf4-43b4-a67d-fbe826c7a4dd"], returnItemNames=["COUNT(*)"], _rows=1, _db_hits=0)
==> EagerAggregation(keys=[], aggregates=["(  INTERNAL_AGGREGATE01b2d10b-7bf4-43b4-a67d-fbe826c7a4dd,CountStar())"], _rows=1, _db_hits=0)
==>   NodeByLabel(label="OwlClass", identifier="c", _rows=146, _db_hits=0)

==> ColumnFilter(symKeys=["  INTERNAL_AGGREGATEe33d15c7-e357-4e32-9f7f-3fbc00dd58f6"], returnItemNames=["COUNT(*)"], _rows=1, _db_hits=0)
==> EagerAggregation(keys=[], aggregates=["(  INTERNAL_AGGREGATEe33d15c7-e357-4e32-9f7f-3fbc00dd58f6,CountStar())"], _rows=1, _db_hits=0)
==>   NodeByLabel(label="OwlClass", identifier="d", _rows=21316, _db_hits=0)
==>     ColumnFilter(symKeys=["c", "n"], returnItemNames=["n"], _rows=146, _db_hits=0)
==>       Extract(symKeys=["c"], exprKeys=["n"], _rows=146, _db_hits=0)
==>         NodeByLabel(label="OwlClass", identifier="c", _rows=146, _db_hits=0)

==> ColumnFilter(symKeys=["count_c", "  INTERNAL_AGGREGATE17be6af7-abbb-4ef7-8250-ff050d6ba7c6"], returnItemNames=["count_c", "count_d"], _rows=1, _db_hits=0)
==> EagerAggregation(keys=["count_c"], aggregates=["(  INTERNAL_AGGREGATE17be6af7-abbb-4ef7-8250-ff050d6ba7c6,Count(d))"], _rows=1, _db_hits=0)
==>   NodeByLabel(label="OwlClass", identifier="d", _rows=146, _db_hits=0)
==>     ColumnFilter(symKeys=["  INTERNAL_AGGREGATE67abdc54-9c44-4353-bf82-e4aff895141e"], returnItemNames=["count_c"], _rows=1, _db_hits=0)
==>       EagerAggregation(keys=[], aggregates=["(  INTERNAL_AGGREGATE67abdc54-9c44-4353-bf82-e4aff895141e,Count(c))"], _rows=1, _db_hits=0)
==>         NodeByLabel(label="OwlClass", identifier="c", _rows=146, _db_hits=0)

-查询3-

MATCH (c:OwlClass) 
RETURN COUNT(*)

MATCH (c:OwlClass) 
WITH null as n 
MATCH (d:OwlClass) 
RETURN COUNT(*)

MATCH (c:OwlClass)
WITH COUNT(c) as count_c
MATCH (d:OwlClass)
RETURN count_c, COUNT(d) as count_d

结果:

COUNT(*)
146
Returned 1 row in 102 ms

COUNT(*)
21316
Returned 1 row in 734 ms

count_c   count_d
146       146
Returned 1 row in 153 ms

执行计划：

==> ColumnFilter(symKeys=["  INTERNAL_AGGREGATE01b2d10b-7bf4-43b4-a67d-fbe826c7a4dd"], returnItemNames=["COUNT(*)"], _rows=1, _db_hits=0)
==> EagerAggregation(keys=[], aggregates=["(  INTERNAL_AGGREGATE01b2d10b-7bf4-43b4-a67d-fbe826c7a4dd,CountStar())"], _rows=1, _db_hits=0)
==>   NodeByLabel(label="OwlClass", identifier="c", _rows=146, _db_hits=0)

==> ColumnFilter(symKeys=["  INTERNAL_AGGREGATEe33d15c7-e357-4e32-9f7f-3fbc00dd58f6"], returnItemNames=["COUNT(*)"], _rows=1, _db_hits=0)
==> EagerAggregation(keys=[], aggregates=["(  INTERNAL_AGGREGATEe33d15c7-e357-4e32-9f7f-3fbc00dd58f6,CountStar())"], _rows=1, _db_hits=0)
==>   NodeByLabel(label="OwlClass", identifier="d", _rows=21316, _db_hits=0)
==>     ColumnFilter(symKeys=["c", "n"], returnItemNames=["n"], _rows=146, _db_hits=0)
==>       Extract(symKeys=["c"], exprKeys=["n"], _rows=146, _db_hits=0)
==>         NodeByLabel(label="OwlClass", identifier="c", _rows=146, _db_hits=0)

==> ColumnFilter(symKeys=["count_c", "  INTERNAL_AGGREGATE17be6af7-abbb-4ef7-8250-ff050d6ba7c6"], returnItemNames=["count_c", "count_d"], _rows=1, _db_hits=0)
==> EagerAggregation(keys=["count_c"], aggregates=["(  INTERNAL_AGGREGATE17be6af7-abbb-4ef7-8250-ff050d6ba7c6,Count(d))"], _rows=1, _db_hits=0)
==>   NodeByLabel(label="OwlClass", identifier="d", _rows=146, _db_hits=0)
==>     ColumnFilter(symKeys=["  INTERNAL_AGGREGATE67abdc54-9c44-4353-bf82-e4aff895141e"], returnItemNames=["count_c"], _rows=1, _db_hits=0)
==>       EagerAggregation(keys=[], aggregates=["(  INTERNAL_AGGREGATE67abdc54-9c44-4353-bf82-e4aff895141e,Count(c))"], _rows=1, _db_hits=0)
==>         NodeByLabel(label="OwlClass", identifier="c", _rows=146, _db_hits=0)

--

自下而上阅读执行计划。本例中的相关部分是行数如何变化。比较查询2和查询3的粗略解释

查询2

首先获取标签为

OwlClass

的所有节点。现在有146个匹配的模式

通过将n设置为null来继续每个模式

通过使用label

OwlClass

获取所有节点，继续每个模式。现在有146*146=21316个匹配模式

数一数，把这些加起来。现在有一种模式（计数）。归还它 查询3

首先获取标签为

OwlClass

的所有节点。现在有146个匹配的模式

数一数，把这些加起来。现在有一种模式（计数）

通过使用label

OwlClass

获取所有节点，继续每个模式。现在有146个匹配的模式

数一数，把这些加起来。现在有一种模式（计数）。归还它

---

你的意思是用r表示

，而不是用null表示none

表示

？不，我不想把任何东西带到第二个查询中。你的第一个语句非常可疑，你真的没有这个关系连接到的任何起点或标签吗？它是简化的测试模型（5k节点，2k：关系的waitsFor）.你能简单地解释一下最后一个执行计划吗？请从头至尾阅读。首先获取标签为
OwlClass
的所有节点。现在有146个匹配的模式。通过将n设置为null来继续每个模式。通过使用label
OwlClass
获取所有节点，继续每个模式。现在有146*146=21316个匹配模式。数一数，把这些加起来。现在有一种模式（计数）。返回它。如果为null，则由于none不会更改基数，因此必须将结果折叠/聚合为较小的基数