neovis.js


Graph visualizations powered by vis.js with data from Neo4j.

Features
Install
Neovis.js can be installed via npm:
npm install --save neovis.js
you can also obtain neovis.js via CDN:
CDN
For ease of use Neovis.js can be obtained from Neo4jLabs CDN:
Most recent release
<script src="https://unpkg.com/neovis.js@2.0.2"></script>
Version without neo4j-driver dependency
<script src="https://unpkg.com/neovis.js@2.0.2/dist/neovis-without-dependencies.js"></script>
Quickstart Example
Let’s go through the steps to reproduce this visualization:

Prepare Neo4j
Start with a blank Neo4j instance, or spin up a blank Neo4j Sandbox. We’ll load the Game of
Thrones dataset, run:
LOAD CSV WITH HEADERS FROM 'https://raw.githubusercontent.com/mathbeveridge/asoiaf/master/data/asoiaf-all-edges.csv'
AS row
MERGE (src:Character {name: row.Source})
MERGE (tgt:Character {name: row.Target})
MERGE (src)-[r:INTERACTS]->(tgt)
ON CREATE SET r.weight = toInteger(row.weight)
We’ve pre-calculated PageRank and ran a community detection algorithm to assign community ids for each Character. Let’s
load those next:
LOAD CSV WITH HEADERS FROM 'https://raw.githubusercontent.com/johnymontana/neovis.js/master/examples/data/got-centralities.csv'
AS row
MATCH (c:Character {name: row.name})
SET c.community = toInteger(row.community),
c.pagerank = toFloat(row.pagerank)
Our graph now consists of Character
nodes that are connected by an INTERACTS
relationships. We can visualize the
whole graph in Neo4j Browser by running:
MATCH p = (:Character)-[:INTERACTS]->(:Character)
RETURN p

We can see characters that are connected and with the help of the force directed layout we can begin to see clusters in
the graph. However, we want to visualize the centralities (PageRank) and community detection results that we also
imported.
Specifically we would like:
- Node size to be proportional to the Character’s
pagerank
score. This will allow us to quickly identify important
nodes in the network.
- Node color to determined by the
community
property. This will allow us to visualize clusters.
- Relationship thickeness should be proportional to the
weight
property on the INTERACTS
relationship.
Neovis.js, by combining the JavaScript driver for Neo4j and the vis.js visualization library will allow us to build this
visualization.
index.html
Create a new html file:
<!doctype html>
<html>
<head>
<title>Neovis.js Simple Example</title>
<style type="text/css">
html, body {
font: 16pt arial;
}
#viz {
width: 900px;
height: 700px;
border: 1px solid lightgray;
font: 22pt arial;
}
</style>
</head>
<body onload="draw()">
<div id="viz"></div>
</body>
</html>
We define some basic CSS to specify the boundaries of a div
and then create a single div
in the body. We also
specify onload="draw()"
so that the draw()
function is called as soon as the body is loaded.
We need to pull in neovis.js
:
<script src="https://unpkg.com/neovis.js@2.0.2"></script>
And define our draw() function:
<script type="text/javascript">
let neoViz;
function draw() {
const config = {
containerId: "viz",
neo4j: {
serverUrl: "bolt://localhost:7687",
serverUser: "neo4j",
serverPassword: "sorts-swims-burglaries",
},
labels: {
Character: {
label: "name",
value: "pagerank",
group: "community",
[NeoVis.NEOVIS_ADVANCED_CONFIG]: {
function: {
title: (node) => viz.nodeToHtml(node, [
"name",
"pagerank"
])
}
}
}
},
relationships: {
INTERACTS: {
value: "weight"
}
},
initialCypher: "MATCH (n)-[r:INTERACTS]->(m) RETURN *"
};
neoViz = new NeoVis.default(config);
neoViz.render();
}
</script>
This function creates a config
object that specifies how to connect to Neo4j, what data to fetch, and how to configure
the visualization.

See simple-example.html for the full code.
module usage
you can also use it as module, but it would require you have a way to import css files
import NeoVis from 'neovis.js';
or you can import the version with bundled dependency
import NeoVis from 'neovis.js/dist/neovis.js';
Api Reference
Api Reference
Build
This project uses git submodules to include the dependencies for neo4j-driver and vis.js. This project uses webpack to
build a bundle that includes all project dependencies. webpack.config.js
contains the configuration for webpack. After
cloning the repo:
npm install
npm run build
npm run typedoc
will build dist/neovis.js
and dist/neovis-without-dependencies.js
neovis.js
Graph visualizations powered by vis.js with data from Neo4j.
Features
Install
Neovis.js can be installed via npm:
you can also obtain neovis.js via CDN:
CDN
For ease of use Neovis.js can be obtained from Neo4jLabs CDN:
Most recent release
Version without neo4j-driver dependency
Quickstart Example
Let’s go through the steps to reproduce this visualization:
Prepare Neo4j
Start with a blank Neo4j instance, or spin up a blank Neo4j Sandbox. We’ll load the Game of Thrones dataset, run:
We’ve pre-calculated PageRank and ran a community detection algorithm to assign community ids for each Character. Let’s load those next:
Our graph now consists of
Character
nodes that are connected by anINTERACTS
relationships. We can visualize the whole graph in Neo4j Browser by running:We can see characters that are connected and with the help of the force directed layout we can begin to see clusters in the graph. However, we want to visualize the centralities (PageRank) and community detection results that we also imported.
Specifically we would like:
pagerank
score. This will allow us to quickly identify important nodes in the network.community
property. This will allow us to visualize clusters.weight
property on theINTERACTS
relationship.Neovis.js, by combining the JavaScript driver for Neo4j and the vis.js visualization library will allow us to build this visualization.
index.html
Create a new html file:
We define some basic CSS to specify the boundaries of a
div
and then create a singlediv
in the body. We also specifyonload="draw()"
so that thedraw()
function is called as soon as the body is loaded.We need to pull in
neovis.js
:And define our draw() function:
This function creates a
config
object that specifies how to connect to Neo4j, what data to fetch, and how to configure the visualization.See simple-example.html for the full code.
module usage
you can also use it as module, but it would require you have a way to import css files
or you can import the version with bundled dependency
Api Reference
Api Reference
Build
This project uses git submodules to include the dependencies for neo4j-driver and vis.js. This project uses webpack to build a bundle that includes all project dependencies.
webpack.config.js
contains the configuration for webpack. After cloning the repo:will build
dist/neovis.js
anddist/neovis-without-dependencies.js