如何将边的二部列表转换为 NetworkX 中的图，每组节点都具有重叠标签？

Question

I'm currently trying to convert a JSON file I have into a bipartite graph using NetworkX. My data looks like this:

    'edges': [
   {'A': 8, 'B': 6},
   {'A': 21, 'B': 4},
   {'A': 27, 'B': 6},
   {'A': 4, 'B': 5},
   ...
   ]

Basically, I have two sets of nodes: A and B. The list of edges than describes how a node of set A connects with a node of set B.

My issue is that the nodes have overlapping identifiers. In other words, I have nodes 0, 1, 2, etc. in set A, and 0, 1, 2, etc. in B (not necessarily the same number of nodes, but their values overlap). When I try to add all the nodes into my graph, I don't get all of the nodes, since some are repeated.

How can I deal with having repeat nodes in my two sets? They are actually different nodes, but NetworkX isn't treating them as such. I saw that you can set a "value" to a node, but I'm not sure how to fit everything together.

In the end, I would like to have a way to generate the nodes such that I can add edges to my graph in the following way:

import networkx as nx

G = nx.Graph()
# Add nodes here

# Add edges
G.add_edges_from(
    (edge['B'], edge['A'])
    for edge in graph['edges']
)

Answer 1

You can add nodes to graph with ID's equal not to values in your dicts, but equal to AB-type plus values:

import networkx as nx

# Create edges dict and graph
edges = [
   {'A': 8, 'B': 6},
   {'A': 21, 'B': 4},
   {'A': 27, 'B': 6},
   {'A': 4, 'B': 5},
]
G = nx.Graph()

for edge in edges:

    # Create A-id
    a_id = 'A' + str(edge['A'])

    # Add node with A-id to graph
    G.add_node(a_id)

    # Add value to node data
    G.nodes[a_id]['value'] = edge['A']

    # Create B-id
    b_id = 'B' + str(edge['B'])

    # Add node with B-id to graph
    G.add_node(b_id)

    # Add value to node data
    G.nodes[b_id]['value'] = edge['B']

    # Add A-id - B-id edge to graph
    G.add_edge(a_id, b_id)

If you will draw the graph:

nx.draw(
    G,
    node_color='#FF0000',
    with_labels=True
)

You will get something like this: