SmallGraph is a simple graph query language for property graph model, i.e. labeled multi-graph each of whose vertex/edge has associated key-value pairs. SmallGraphs and GraphD are the key components of the toolchain that implements the effective usage of this query language.

First, download SmallGraphs/GraphD and place it where you can easily access it. Assuming ~/bin is already in your PATH environment:

curl -RLO https://github.com/downloads/netj/SmallGraphs/graphd-latest.sh

mv graphd-latest.sh ~/bin/graphd
chmod +x ~/bin/graphd

Please note that node.js is essential to running SmallGraphs/GraphD. Make sure it is installed to your system.

Now, you need to decide where you want to keep all your GraphD metadata and some actual data. The following instructions will assume graphs/ is the directory you chose, and already present at that directory.

mkdir -p graphs
cd graphs

To start the GraphD server at the current working directory, run:

graphd start

(You can optionally pass a PortNumber if you want to use a port other than the default 53411.)

Before sketching or running any query, you need to tell GraphD where your data is, and how it is laid out there. You can use different backends for different graphs, and instructions will vary depending on which you choose.

To use data in you MySQL server and view it as a graph, run:

graphd create mysql NameOfTheGraph MySQLDatabaseName Username Password

(You can pass optional arguments Hostname and PortNumber at the end if they differ from localhost and 3306.)

After running it, you will find an rdbLayout.json file generated, which contains how the vertices and edges are laid out in your relational database. You can modify this JSON file to fix the labels, or remove unwanted vertex and edge types.

Don't forget to add to PATH where MySQL is installed. For instance:

PATH="/usr/local/mysql/bin:$PATH"

Otherwise, you may get errors, such as mysqldump: not available.

Giraph is an open-source implementation of Pregel that runs on a Hadoop cluster. Pregel is a distributed graph processing style used by Google, which allows you to describe graph algorithms by specifying how each vertex should process and transmit messages.

To use Giraph for processing our queries, run:

graphd create giraph NameOfTheGraph

Then, move into the graph, and import some RDF NTriples:

cd NameOfTheGraph
graphd import GraphDataFile.nt

This step will dictionary encode all the triples, i.e. assign unique number to each URI and automatically derive a graph schema for you based on what types of vertices happen to be linked by edges of which types.

You can now open http://localhost:53411/ from your web browser to sketch and run graph queries.

(WebKit-based browsers, such as Safari or Chrome are recommended.)

When you are done with using SmallGraphs and GraphD, simply run the following command from where you started it:

graphd stop

The remaining sections will explain build instructions and organization of this source tree.

SmallGraphs, GraphD, and other tools here are written mostly in CoffeeScript or Bash, and requires node.js for building and running. Please install at least the following dependencies to your system before you proceed.

• node.js
• Git
• GNU Make
• GNU Bash, coreutils, sed, awk, grep, find, tar, gzip
• JDK 6+
• Apache Maven 3+

On Mac OS X, you are recommended to use Homebrew to install most of the missing ones.

brew install node git coreutils maven

On Debian-based systems, you can install most of them by running:

sudo apt-get install git build-essentials coreutils findutils bash sed gawk tar maven

You may need to run the following git commands, to make sure you have all the required submodules checked out:

git submodule init
git submodule update

To build, simply run:

make

Everything built will be staged under @prefix@/. You can add a directory to your PATH environment for convenience.

PATH="$PWD/@prefix@/bin:$PATH"

graphd ...

If you want to install it to your system or in your home, say ~/smallgraphs/bin, then run:

make install PREFIX=~/smallgraphs

PATH="~/smallgraphs/bin:$PATH"

graphd ...

Another option is to create a self-extracting executable by running:

make PACKAGEEXECUTES=bin/graphd

This will generate a flat executable file graphd-*.sh. You can then put this single file anywhere you want and use it as any other script or executable.

• ui/ contains code related to the frontend user interface that lets you sketch queries and browse through results.
• graphd/ contains the backend code that compiles queries and drives other systems to run them.
• shell/ contains the command-line interface code.
• smallgraph/ contains the parser and serializer for our SmallGraph DSL.
• tools/ contains some useful tools for handling graph data.
• doc/ contains some documentation.

• SmallGraphs and GraphD were initialy created by Jaeho Shin as a final project for the Data Visualization course at Stanford University. You can find a short paper about it with more information from its wiki page.

• Most part of the software is written in CoffeeScript and Bash with several commonly used Unix tools.

• The frontend UI, SmallGraphs works with a WebKit-based browser, such as Safari or Chrome. It heavily depends on jQuery and jQuery UI. jQuery-SVG and jQuery-cookie made it possible to keep the code short. d3.js also played an important role in visualizing the results.

• The backend GraphD server is built using express on top of node.js, and connects to MySQL databases with node-mysql.

• SmallGraph parser is generated with Jison.