This is a tool for the use of modular sturcture to find related items within complex networks based on a small number of seed nodes.
The contents of this repository includes code for the research paper:
From clusters to queries: exploiting the uncertainty in the modularity landscape of complex networks James P Gilbert and Jamie Twycross MLG at KDD 2018.
Written for Python version 3.5+
To install, create a virtualenv environment and run
$ python setup.py install
Currently, the API is a little involved to work with. However, here is a basic example of how to use the tool.
from cluster_query_tool.indexer import get_index
from cluster_query_tool.louvain_consensus import query_vector, membership_matrix
import networkx as nx
graph = nx.read_edgelist("data/eu_email/network.txt")
# This may take some time but
index = get_index(graph)
memberships, nmap = memership_matrix(graph, index)
# initial seed nodes
query = [1, 2, 5 ,8]
seed_set = np.array([nmap[x] for x in query])
vec = query_vector(memberships, seed_set)
vec is an np array of values between 0 and 1 for the membership of node i
(vec[nmap[i]] is the value for index of node i).