jamesjgardner / gzero

GZero simplifies large scale graph-based computing, storage, and machine learning model predictions.

gzero

GZero simplifies large scale graph-based computing, storage, and machine learning model predictions. All features and labels will be expressed as graph traversals. A user may specify the features necessary to construct a feature vector with the appropriate labels and the framework will provide an easy API for training, evaluating, and performing predictions with stored models.

Getting Started (Local Development)

Titan 1.1

Clone titan branch titan11 and run:
mvn clean install -DskipTests=true -Paurelius-release -Dgpg.skip=true

Cassandra & Elastic Search

Titan 1.1 comes with cassandra and elasticsearch. Start cassandra and enable thrift, then start elasticsearch:

titan/bin/cassandra -f  
titan/bin/nodetool enablethrift  
titan/bin/elasticsearch

Gremlin Server

Ensure config in titan/conf/gremlin-server.yaml is using the HttpChannelizer to enable the REST API. You'll also want to make sure the graph properties include cassandra and elastic search. You can find an example in conf/gremlin-server.yaml

Now, start the server: bin/gremlin-server.sh

Running GZero API

Rename src/main/resources/application.conf and src/main/resources/logback.xml and configure to suit your needs

Execute

sbt compile
sbt run

to start the API on port 8080.

Building for Linux

Install Titan

Follow the steps above to build Titan11 branch

Configure

Create configuration files that fit your needs in src/main/resources

sbt

A debian package will be generated in the target folder that can be installed.

sbt compile
sbt publish
sbt debian:packageBin
sbt debian:publish
sudo dpkg -i target/gzero_0.0.1-SNAPSHOT_all.deb

You can now start/stop the service with sudo service gzero stop|start|restart

API

POST /edge

Create an edge. Will create verticies if they do not exist.

Request:

curl \
	-H 'Content-Type: application/json' \
	-X POST localhost:8080/edge \
	-d '{"label":"drove", "head":{"label":"person","properties" : {"name":"Bonnie"}},"tail":{"label":"vehicle", "properties":{"name":"1932 Ford V-8 B-400 convertible sedan"}}}'

POST /vertex

Create a vertex.

Request:

curl \
	-H 'Content-Type: application/json' \
	-X POST localhost:8080/vertex \
	-d '{"label":"person","properties":{"name":"Clyde"}}'

GET /query

Run a gremlin query against the graph.

Request:

curl \
	-H 'Content-Type: application/json' \
	-X GET localhost:8080/query \
	-d '{"gremlin":"g.V().has(\"name\", \"Bonnie\")"}'

Code

If you are interested in contributing and jumping into the code, start with GZeroService.scala