CODEWITHSUNDEEP
pythontwitternetworkxgraph-theory
eeno
eeno

Reputation: 45

Is there a way to find a bidirectional relationship between nodes in NetworkX?

I have a network of Twitter users and their followers which I am modelling using NetworkX. I am trying to find a bidirectional link between users, i.e. if one node follows one of its neighbors, does that neighbor also follow that node.

Is there a built in function within NetworkX that would accomplish this? I tried using nx.reciprocity() but it just returns a single value rather than a dictionary.

Upvotes: 3

Views: 2490

Answers (1)

Christopher Peisert
Christopher Peisert

Reputation: 24104

You can determine if there is an edge connection between two nodes with networkx.Graph.has_edge. This could then be used to test if there are opposite directed edges between nodes.

import networkx as nx

def have_bidirectional_relationship(G, node1, node2):
    return G.has_edge(node1, node2) and G.has_edge(node2, node1)


G = nx.DiGraph()
G.add_edge(1, 2)
G.add_edge(2, 1)
G.add_edge(3, 4)

print(f"Nodes 1, 2 have opposite edges: {have_bidirectional_relationship(G, 1, 2)}")
print(f"Nodes 3, 4 have opposite edges: {have_bidirectional_relationship(G, 3, 4)}")

Output

Nodes 1, 2 have opposite edges: True
Nodes 3, 4 have opposite edges: False

Find all nodes with bidirectional relationships

biconnections = set()
for u, v in G.edges():
    if u > v:  # Avoid duplicates, such as (1, 2) and (2, 1)
        v, u = u, v
    if have_bidirectional_relationship(G, u, v):
        biconnections.add((u, v))

print(biconnections)

Output

{(1, 2)}

Upvotes: 2

Related Questions