This is a package for graphs processing and analytics on scale. It is built on top of Apache Spark and relies on DataFrame abstraction. Users can write highly expressive queries by leveraging the DataFrame API, combined with a new API for network motif finding. The user also benefits from DataFrame performance optimizations within the Spark SQL engine. GraphFrames works in Java, Scala, and Python.

You can find user guide and API docs at https://graphframes.io

This project was in maintenance mode for some time, but we are happy to announce that it is now back in active development!

GraphFrames scala core and Spark-Connect plugin are published in the Sonatype Central. Namespace is io.graphframes.

# Interactive Scala/Java
# For Spark 3.5.x, scala 2.12
$ spark-shell --packages io.graphframes:graphframes-spark3_2.12:0.9.2
# For Spark 3.5.x, scala 2.13
$ spark-shell --packages io.graphframes:graphframes-spark3_2.13:0.9.2
# For Spark 4.0.x
$ spark-shell --packages io.graphframes:graphframes-spark4_2.13:0.9.2
# Interactive Python, Spark 3.5.x
$ pyspark --packages io.graphframes:graphframes-spark3_2.12:0.9.2
# Interactive Python, Spark 4.0.x
$ pyspark --packages io.graphframes:graphframes-spark4_2.13:0.9.2

Python API is published in the PyPi:

pip install graphframes-py

NOTE! Python distribution does not include JVM-core. You need to add it to your cluster or Spark-Connect server!

To add GraphFrames to your spark connect server, you need to specify the plugin name, for example:

./sbin/start-connect-server.sh \
  --conf spark.connect.extensions.relation.classes=org.apache.spark.sql.graphframes.GraphFramesConnect \
  --packages io.graphframes.graphframes-connect-spark4_2.13:0.9.2
  --conf spark.checkpoint.dir=${CHECKPOINT_DIR}

NOTE! GraphFrames is relying on iterative graph algorithms and uses checkpoints internally to avoid infinite growing of the Spark's Logical Plan. Spark-Connect API does not provide the way to specify the checkpoint dir and it should be specified via spark.checkpoint.dir configuration!

Now you can create a GraphFrame as follows.

In Python:

from pyspark.sql import SparkSession
from graphframes import GraphFrame
spark = SparkSession.builder.getOrCreate()
nodes = [
    (1, "Alice", 30),
    (2, "Bob", 25),
    (3, "Charlie", 35)
]
nodes_df = spark.createDataFrame(nodes, ["id", "name", "age"])
edges = [
    (1, 2, "friend"),
    (2, 1, "friend"),
    (2, 3, "friend"),
    (3, 2, "enemy")  # eek!
]
edges_df = spark.createDataFrame(edges, ["src", "dst", "relationship"])
g = GraphFrame(nodes_df, edges_df)

Now let's run some graph algorithms at scale!

g.inDegrees.show()
# +---+--------+
# | id|inDegree|
# +---+--------+
# |  2|       2|
# |  1|       1|
# |  3|       1|
# +---+--------+
g.outDegrees.show()
# +---+---------+
# | id|outDegree|
# +---+---------+
# |  1|        1|
# |  2|        2|
# |  3|        1|
# +---+---------+
g.degrees.show()
# +---+------+
# | id|degree|
# +---+------+
# |  1|     2|
# |  2|     4|
# |  3|     2|
# +---+------+
g2 = g.pageRank(resetProbability=0.15, tol=0.01)
g2.vertices.show()
# +---+-----+---+------------------+
# | id| name|age|          pagerank|
# +---+-----+---+------------------+
# |  1| John| 30|0.7758750474847483|
# |  2|Alice| 25|1.4482499050305027|
# |  3|  Bob| 35|0.7758750474847483|
# +---+-----+---+------------------+
# GraphFrames' most used feature...
# Connected components can do big data entity resolution on billions or even trillions of records!
# First connect records with a similarity metric, then run connectedComponents.
# This gives you groups of identical records, which you then link by same_as edges or merge into list-based master records.
sc.setCheckpointDir("/tmp/graphframes-example-connected-components")  # required by GraphFrames.connectedComponents
g.connectedComponents().show()
# +---+-----+---+---------+
# | id| name|age|component|
# +---+-----+---+---------+
# |  1| John| 30|        1|
# |  2|Alice| 25|        1|
# |  3|  Bob| 35|        1|
# +---+-----+---+---------+
# Find frenemies with network motif finding! See how graph and relational queries are combined?
(
    g.find("(a)-[e]->(b); (b)-[e2]->(a)")
    .filter("e.relationship = 'friend' and e2.relationship = 'enemy'")
    .show()
)
# These are paths, which you can aggregate and count to find complex patterns.
# +------------+--------------+----------------+-------------+
# |           a|             e|               b|           e2|
# +------------+--------------+----------------+-------------+
# |{2, Bob, 25}|{2, 3, friend}|{3, Charlie, 35}|{3, 2, enemy}|
# +------------+--------------+----------------+-------------+

To learn more about GraphFrames, check out these resources:

• GraphFrames Documentation
• GraphFrames Network Motif Finding Tutorial
• Introducing GraphFrames
• On-Time Flight Performance with GraphFrames for Apache Spark

• GraphFrames Google Group
• #graphframes Discord Channel on GraphGeeks
• Graph Operations in Apache Spark Using GraphFrames
• Executing Graph Algorithms with GraphFrames on Databricks

graphframes-py is our Official PyPi Package

We recommend using the Spark Packages system to install the latest version of GraphFrames, but now publish a build of our Python package to PyPi in the graphframes-py package. It can be used to provide type hints in IDEs, but does not load the java-side of GraphFrames so will not work without loading the GraphFrames package. See Installation and Quick-Start.

pip install graphframes-py

WARNING!

This project does not own or control the graphframes PyPI package (installs 0.6.0) or graphframes-latest PyPI package (installs 0.8.4).

WARNING!

Maven:

<dependencies>
    <dependency>
        <groupId>io.graphframes</groupId>
        <artifactId>graphframes-spark4_2.13</artifactId>
        <version>0.9.2</version>
    </dependency>
</dependencies>

SBT:

libraryDependencies += "io.graphframes" %% "graphframes-spark4" % "0.9.2"

WARNING!

=========================

Due to governance problems and limitations, all the new releases of GraphFrames will be published to the Maven Central under the namespace io.graphframes (not org.graphframes)!

=========================

To learn how GraphFrames works internally to combine graph and relational queries, check out the paper GraphFrames: An Integrated API for Mixing Graph and Relational Queries, Dave et al. 2016.

To compile the core project, run build/sbt package from the project home directory. To compile the Spark Connect Plugin, run build/sbt connect/package

This project is compatible with Spark 3.5.x and Spark 4.0.x. Significant speed improvements have been made to DataFrames in recent versions of Spark, so you may see speedups from using the latest Spark version.

GraphFrames was made as collaborative effort among UC Berkeley, MIT, Databricks and the open source community. At the moment GraphFrames is maintained by the group of individual contributors.

See contribution guide

See release notes.

GraphFrames project is publishing SNAPSHOTS (nightly builds) to the "Central Portal Snapshots." Please read this section of the Sonatype documentation to check how can you use snapshots in your project.

GroupId: io.graphframes ArtifactIds:

• graphframes-spark3_2.12
• graphframes-spark3_2.13
• graphframes-connect-spark3_2.12
• graphframes-connect-spark3_2.13
• graphframes-spark4_2.13
• graphframes-connect-spark4_2.13