apoc.meta.data

Procedure

apoc.meta.data(config MAP<STRING, ANY>) - examines the full graph and returns a table of metadata.

This procedure is not considered safe to run from multiple threads. It is therefore not supported by the parallel runtime (introduced in Neo4j 5.13). For more information, see the Cypher Manual → Parallel runtime.

Signature

apoc.meta.data(config = {} :: MAP) :: (label :: STRING, property :: STRING, count :: INTEGER, unique :: BOOLEAN, index :: BOOLEAN, existence :: BOOLEAN, type :: STRING, array :: BOOLEAN, sample :: LIST<ANY>, left :: INTEGER, right :: INTEGER, other :: LIST<STRING>, otherLabels :: LIST<STRING>, elementType :: STRING)

Input parameters

Name Type Default

config

MAP

{}

Config parameters

This procedure supports the following config parameters:

Table 1. Config parameters
Name Type Default Description

sample

INTEGER

1000

Number of nodes to sample. Setting sample to -1 will remove sampling.

Sampling

Specify the sample parameter (1000 by default) to analyze a subset of the data.

The sample, along with the count of nodes for each label, is used to calculate a skip value. Since this value is generated using a random number generator, results obtained through the sampling method may vary between subsequent runs.

Example 1. Calculating skip count for data sampling

If a database contains 500 nodes with the label Foo label, the skip count for that label is calculated as follows:

The skip count per node label is determined by generating a random number between (totalNodesForLabel / sample) ± 0.1.

Sample 10: skipCount = 500 / 10 = 50
The resulting skip count will be between 45 and 55.

Sample 50: skipCount = 500 / 50 = 10
The resulting skip count will be between 9 and 11.

Sample 100: skipCount = 500 / 100 = 5
The resulting skip count will be 5.

The skip count represents the number of nodes skipped before one is examined. For instance, with a skip count of 5, every 5th node is examined. Consequently, a higher sample number results in more nodes being sampled.

To stop sampling set sample: -1.

Output parameters

Name Type

label

STRING

property

STRING

count

INTEGER

unique

BOOLEAN

index

BOOLEAN

existence

BOOLEAN

type

STRING

array

BOOLEAN

sample

LIST<ANY>

left

INTEGER

right

INTEGER

other

LIST<STRING>

otherLabels

LIST<STRING>

elementType

STRING

Usage Examples

The examples in this section are based on the following sample graph:

CREATE (Keanu:Person {name:'Keanu Reeves', born:1964})
CREATE (TomH:Person {name:'Tom Hanks', born:1956})
CREATE (Carrie:Person {name:'Carrie-Anne Moss', born:1967})
CREATE (Laurence:Person {name:'Laurence Fishburne', born:1961})
CREATE (Hugo:Person {name:'Hugo Weaving', born:1960})
CREATE (LillyW:Person {name:'Lilly Wachowski', born:1967})
CREATE (LanaW:Person {name:'Lana Wachowski', born:1965})

CREATE (TheMatrix:Movie {title:'The Matrix', released:1999, tagline:'Welcome to the Real World'})
CREATE (TheMatrixReloaded:Movie {title:'The Matrix Reloaded', released:2003, tagline:'Free your mind'})
CREATE (TheMatrixRevolutions:Movie {title:'The Matrix Revolutions', released:2003, tagline:'Everything that has a beginning has an end'})
CREATE (SomethingsGottaGive:Movie {title:"Something's Gotta Give", released:2003})
CREATE (TheDevilsAdvocate:Movie {title:"The Devil's Advocate", released:1997, tagline:'Evil has its winning ways'})
CREATE (Something:Something:Else {foo: 'bar'})

CREATE (YouveGotMail:Movie {title:"You've Got Mail", released:1998, tagline:'At odds in life... in love on-line.'})
CREATE (SleeplessInSeattle:Movie {title:'Sleepless in Seattle', released:1993, tagline:'What if someone you never met, someone you never saw, someone you never knew was the only someone for you?'})
CREATE (ThatThingYouDo:Movie {title:'That Thing You Do', released:1996, tagline:'In every life there comes a time when that thing you dream becomes that thing you do'})
CREATE (CloudAtlas:Movie {title:'Cloud Atlas', released:2012, tagline:'Everything is connected'})

CREATE (Keanu)-[:ACTED_IN {roles:['Neo']}]->(TheMatrix)
CREATE (Keanu)-[:ACTED_IN {roles:['Neo']}]->(TheMatrixReloaded)
CREATE (Keanu)-[:ACTED_IN {roles:['Neo']}]->(TheMatrixRevolutions)
CREATE (Keanu)-[:ACTED_IN {roles:['Julian Mercer']}]->(SomethingsGottaGive)
CREATE (Keanu)-[:ACTED_IN {roles:['Kevin Lomax']}]->(TheDevilsAdvocate)

CREATE (TomH)-[:ACTED_IN {roles:['Joe Fox']}]->(YouveGotMail)
CREATE (TomH)-[:ACTED_IN {roles:['Sam Baldwin']}]->(SleeplessInSeattle)
CREATE (TomH)-[:ACTED_IN {roles:['Mr. White']}]->(ThatThingYouDo)
CREATE (TomH)-[:ACTED_IN {roles:['Zachry', 'Dr. Henry Goose', 'Isaac Sachs', 'Dermot Hoggins']}]->(CloudAtlas)

CREATE (Keanu)-[:LIKES {rate:10}]->(Carrie)
CREATE (Keanu)-[:LIKES {rate:6}]->(TomH)
CREATE (Keanu)-[:LIKES {rate:4}]->(Laurence)
CREATE (Keanu)-[:LIKES {rate:8}]->(Hugo)
CREATE (Keanu)-[:LIKES {rate:9}]->(LillyW)
CREATE (Keanu)-[:LIKES {rate:6}]->(LanaW)
CREATE (Keanu)-[:LIKES {rate:100}]->(TheMatrix)
CREATE (Carrie)-[:LIKES {rate:7}]->(Keanu)

CREATE (Something)-[:RELATED_TO {rate:99}]->(TheMatrix)
CREATE (Keanu)-[:RELATED_TO {rate:12}]->(TheMatrix)
CREATE (Keanu)-[:RELATED_TO {rate:12}]->(TheMatrixReloaded)
CREATE (Keanu)-[:RELATED_TO {rate:23}]->(TheMatrixRevolutions)
CREATE (Carrie)-[:RELATED_TO {rate:34}]->(TheMatrix)
CREATE (TheMatrix)-[:RELATED_TO {rate:34}]->(Laurence)
CREATE (TheMatrixReloaded)-[:RELATED_TO {rate:345}]->(LanaW)
CALL apoc.meta.data();
Table 2. Results
label property count unique index existence type array sample left right other otherLabels elementType

"ACTED_IN"

"Person"

2

false

false

false

"RELATIONSHIP"

true

null

4

0

["Movie"]

[]

"relationship"

"ACTED_IN"

"roles"

0

false

false

false

"LIST"

true

null

0

0

[]

[]

"relationship"

"LIKES"

"Person"

2

false

false

false

"RELATIONSHIP"

true

null

4

1

["Movie", "Person"]

[]

"relationship"

"LIKES"

"rate"

0

false

false

false

"INTEGER"

false

null

0

0

[]

[]

"relationship"

"RELATED_TO"

"Person"

2

false

false

false

"RELATIONSHIP"

true

null

2

0

["Movie"]

[]

"relationship"

"RELATED_TO"

"rate"

0

false

false

false

"INTEGER"

false

null

0

0

[]

[]

"relationship"

"RELATED_TO"

"Movie"

2

false

false

false

"RELATIONSHIP"

false

null

1

2

["Person"]

[]

"relationship"

"RELATED_TO"

"Something"

1

false

false

false

"RELATIONSHIP"

false

null

1

0

["Movie"]

[]

"relationship"

"RELATED_TO"

"Else"

1

false

false

false

"RELATIONSHIP"

false

null

1

0

["Movie"]

[]

"relationship"

"Person"

"RELATED_TO"

2

false

false

false

"RELATIONSHIP"

true

null

2

0

["Movie"]

[]

"node"

"Person"

"ACTED_IN"

2

false

false

false

"RELATIONSHIP"

true

null

4

0

["Movie"]

[]

"node"

"Person"

"LIKES"

2

false

false

false

"RELATIONSHIP"

true

null

4

1

["Movie", "Person"]

[]

"node"

"Person"

"born"

0

false

false

false

"INTEGER"

false

null

0

0

[]

[]

"node"

"Person"

"name"

0

false

false

false

"STRING"

false

null

0

0

[]

[]

"node"

"Movie"

"RELATED_TO"

2

false

false

false

"RELATIONSHIP"

false

null

1

2

["Person"]

[]

"node"

"Movie"

"title"

0

false

false

false

"STRING"

false

null

0

0

[]

[]

"node"

"Movie"

"tagline"

0

false

false

false

"STRING"

false

null

0

0

[]

[]

"node"

"Movie"

"released"

0

false

false

false

"INTEGER"

false

null

0

0

[]

[]

"node"

"Something"

"RELATED_TO"

1

false

false

false

"RELATIONSHIP"

false

null

1

0

["Movie"]

[]

"node"

"Something"

"foo"

0

false

false

false

"STRING"

false

null

0

0

[]

[]

"node"

"Else"

"RELATED_TO"

1

false

false

false

"RELATIONSHIP"

false

null

1

0

["Movie"]

[]

"node"

"Else"

"foo"

0

false

false

false

"STRING"

false

null

0

0

[]

[]

"node"

The unique column shows if there is an unique constraint in that specific label and property. Similarly, the index column show if there is an index or not, while the existence looks for an existence constraint.

The array column check if the row is of type array and, in case type columns is "RELATIONSHIP", the result will be true if there is at least one node with 2 outgoing relationships with the type of relation given by label or property column, so ACTED_IN in the example above.

The left, right and count columns regard only rows with column type equals to "RELATIONSHIP" (otherwise they are equal to 0). Please note that, because we examine a sample, these counts are just estimates to give an overview and proximity to actual values depends on the dataset and the sample set (default 100).

In particular the count column indicates the number of nodes with an outgoing relationship (e.g. the row with label = ACTED_IN and property = Person has count 2 because there are 2 nodes (node:Person)-[:ACTED_IN]→(), i.e. (Keanu) and (TomH)).

The left value represents the ratio (rounded down) of the count of the outgoing patterns for a certain label and a specific type of relationship to count. In cypher, it corresponds to:

MATCH p=(start:`<LABEL>`)-[:`<TYPE>`]->()
WITH count(distinct start) as nodes, count(p) as counts
RETURN CASE when nodes = 0 then 0 else counts / nodes end

For example, regarding the row with label = RELATED_TO and property = Movie, there are 2 relationships (:Movie)-[rel:RELATED_TO]→(), i.e (TheMatrix)-[:RELATED_TO {rate:34}]→(Laurence) and (TheMatrixReloaded)-[:RELATED_TO {rate:345}]→(LanaW). So the left value is 1 (2 relationships divided by 2 nodes found).

Instead, regarding the row with label = LIKES and property = Person, there are 8 relationships (:Person)-[rel:LIKES]→(), 7 starting from (Keanu) node, and 1 from (Carrie). Then the left value is 4 (8 relationships divided by 2 nodes found).

The right value is the ratio (rounded down) of the count of the incoming patterns for a certain label and a specific type of relationship to count, where the patterns included in the count are those in which there is an equivalent outgoing relationship. In cypher it corresponds to:

MATCH p=(start:`<LABEL>`)<-[:`<TYPE>`]-()
WHERE exists((start:`<LABEL>`)-[:`<TYPE>`]->())
WITH count(distinct start) as nodes, count(p) as counts
RETURN CASE when nodes = 0 then 0 else counts / nodes end

For example, regarding the row with label = RELATED_TO and property = Movie, the (TheMatrix) node, which has an outgoing RELATED_TO relationship, has 3 incoming relationships as well, while the TheMatrixReloaded node has 1 incoming relationship. So the right value is 2, that is 4 divided by 2 node founds.

Therefore, via the right and left values, we provide a dataset estimate of the possible degree averages.