NEAAR

NEtwork Analysis in ARangodb

Installation

To install the package, clone the repository and type on your terminal:

python setup.py install

Or, if you want to install in developer mode

python setup.py develop

Set up

To use the package, is required to have ArangoDB installed.

To access the databases you'll be asked for a password: is sufficient to create a file called config.json on the repository with the structure:

{
 "host" : "https://127.0.0.1:8529",
 "username" : "root",
 "password" : "1234"
}

port 8529 is the default for ArangoDB.

Now everything is set.

Example

The full example on how to use NEAAR is reported in the file pipeline.py in this repository, that describe what to do in case of a multipartite graph.

import os

from arango import ArangoClient # python-arango import.
import networkx as nx

from neaar import pa_utils as pa

host, username, password = pa.load_pass(filename='config.json')

# Initialize the client for ArangoDB. Connect to "_system" database as root user.
client = ArangoClient(hosts=host)
db     = client.db('_system', username=username, password=password)

Now a connection with the database _system is established.

filename = 'data/multipartite.gexf'

# load a graph as multipartite and return a python-arango and a NetworkX graph
Ardb_Net, Nx_Net = pa.read_gexf(db, filename=filename, multipartite=True,
                       				nodes_collection_name='collection',
                       				edges_collection_name='edges',
                       				graph_name='multipartite')

Now we can run queries on the network. An example is the traversal that look for the first neighbours of a starting node:

first_neighbours = pa.traverse(db=db, starting_node='2',
                                       graph_name='multipartite',
                                       direction='any',
                                       item_order='forward',
                                       min_depth=0,
                                       max_depth=1,
                                       vertex_uniqueness='global')

Now the dictionary first_neighbours contains information on nodes and paths crossed by the traversal.

It is possible to extract the subnetwork of selected nodes with the function subgraph of NetworkX, and save all the information

# Extract subgraph
Nx_Sub_Net = Nx_Net.subgraph([vertex['label'] for vertex in astenia_first_neighbours['vertices']])

# Take back added information on Arango documents
for node in Nx_Sub_Net:
  attr = pa.get_vertex(db, {'label':node}, ["collection_0","collection_1","collection_2"])
  nx.set_node_attributes(Nx_Sub_Net, {node : attr})

And re-upload the subnetwork:

pa.multipartite_to_arango(db,Nx_Sub_Net, "subnet","subedge","subgraph")

Authors