Building terms relations within a network

问题

I am trying to represent relationships between numbers in column A and their corresponding values in B.

A            B
Home     [Kitchen, Home, Towel]
Donald   [US, 02 , Donald, Trump]
Trump    [Trump,Family, Cat, Dog]
Dog      [Dog,Cat,Paws]

Numbers in column A and numbers in B are nodes in a graph. I would like to connect elements in B to A or to each other. For example:

Home in A is linked with itself; if I look within B column (the value appears only in the first row), Home in B is connected to Kitchen and Towel (ingoing link);
Donald is linked with itself in as Donald is only in B; however, Donald in B is connected also with US, 02 and Trump (ingoing link);
Trump has an outgoing link with Donald and ingoing links (Family, Cat and Dog);
Dog has an outgoing link with Trump and ingoing links (Cat and Paws).

The rule should be the following then:

if a word in A is in another row in B, then create an outgoing link;
for each word in B create an ingoing link to the word in A, if the word in A is also included in B.

How should I adjust my code?

file = file.assign(B=file.B.map(list)).explode('B')


G = nx.DiGraph()
nx.add_path(G, file['A'])
nx.add_path(G, file['B'])

nx.draw_networkx(G)
plt.show()

回答1:

Converting your table to a pandas dataframe and then looping through its rows you can add the corresponding edges like this:

import networkx as nx
import pandas as pd
from pyvis.network import Network


df = pd.DataFrame(
    [
        ['Home', ['Kitchen', 'Home', 'Towel']],
        ['Donald', ['US', '02' , 'Donald', 'Trump']],
        ['Trump', ['Trump','Family', 'Cat', 'Dog']],
        ['Dog', ['Dog', 'Cat' , 'Paws']]
    ],
    columns=['A', 'B']
)

G = nx.DiGraph()

for i, j in enumerate(df['A']):
    for index, row in df.iterrows():
        if i != index:
            if j in row['B']:
                G.add_edge(row['A'], j)
        else:
            for n in row['B']:
                if j != n:
                    G.add_edge(j, n)

    if G.in_degree(j) == 0:
        G.add_edge(j , j)

N = Network(directed=True)  # using pyvis to show self loops as well

for n, attrs in G.nodes.data():
    N.add_node(n)

for e in G.edges.data():
    N.add_edge(e[0], e[1])

N.write_html('graph.html')

Which gave me the following graph:

Hope this is what you wanted!

来源：https://stackoverflow.com/questions/63441510/building-terms-relations-within-a-network

标签

python

graph

networkx