从networkx python中的重复集合列表生成加权边答案

【问题标题】：Generate weighted edges from duplicate list of set in networkx python从networkx python中的重复集合列表生成加权边
【发布时间】：2022-11-03 21:48:28
【问题描述】：

所以，我想用这样的集合列表中的数据创建加权图：

temp_comb_test = [{'AN', 'TA'}, {'TA', 'DP'},{'AS','TA'},{'HS','AS'},{'HS','TA'},{'TA','AA'},{'LS','TA'}]

那些加权的，是从重复的边缘生成的。使用set() 类型的数据仅仅是因为在集合中，（A，B）和（B，A）是相等/重复的数据（在我的知识列表中，元组不能这样做）。所以我写了一个这样的代码来添加加权边缘：

G_Author = nx.Graph()

temp = [] # to keep temporary relation (no dupe)

for iter1 in temp_comb_test:
    
    if len(temp) == 0: # to add first set
        temp.append(iter1)
        G_Author.add_edges_from([iter1], weight = 1)

        print("A - Data ", iter1, " In")
        print("temp :", temp)
        print(G_Author.edges.data(), '\n') 

    else:
        for iter2 in temp: # iterate temporary list
            if iter1 == iter2: # checking duplicate set

                nod1, nod2 = iter1
                nod3, nod4 = iter2                                          
                
                if ((nod1 == nod3) and (nod2 == nod4) and (nod1 != nod4)): # if set look like --> (A, B) == (A, B)

                    print("F -", "new :", iter1, "old :", iter2)
                    
                    wei = nx.get_edge_attributes(G_Author, "weight")
                    wei2 = wei[nod4, nod3] + 1      # ====> THE PROBLEM 2
                    nx.set_edge_attributes(G_Author, {(nod3, nod4): {"weight": wei2}})
                    print(G_Author.edges.data(), '\n')

                elif ((nod1 != nod3) and (nod2 != nod4) and (nod1 == nod4)): # if duplicate set looks like --> (A, B) == (B, A)
                    print("F -", iter1, iter2)
                    wei3 = nx.get_edge_attributes(G_Author, "weight")
                    wei4 = wei3[nod3, nod4] + 1
                    nx.set_edge_attributes(G_Author, {(nod2, nod1): {"weight": wei4}})
            else:
                nd1, nd2 = iter1
                print("E - Data {", nd1, ",", nd2, "} in || iter1 :", iter1)
                G_Author.add_edge(nd2, nd1, weight = 1)     # ====> THE PROBLEM 1
                temp.append({nd2, nd1})
  
                print("temp :", temp)
                print(G_Author.edges.data(), '\n')

当我运行此代码时，第一个元素已成功添加到图形中。接下来是第一个 else 条件。但是，问题出现在下一次迭代中，如下所示：

A - Data  {'TA', 'AN'}  in
temp : [{'TA', 'AN'}]
[('TA', 'AN', {'weight': 1})] 

E - Data { DP , TA } in || iter1 : {'DP', 'TA'}
temp : [{'TA', 'AN'}, {'DP', 'TA'}]
[('TA', 'AN', {'weight': 1}), ('TA', 'DP', {'weight': 1})] 

F - new : {'DP', 'TA'} old : {'DP', 'TA'}

---------------------------------------------------------------------------
KeyError                                  Traceback (most recent call last)
<ipython-input-136-f103fe75d64d> in <module>
     33                     wei = nx.get_edge_attributes(G_Author, "weight")
     34                     # print(wei)
---> 35                     wei2 = wei[nod3, nod4] + 1
     36                     nx.set_edge_attributes(G_Author, {(nod3, nod4): {"weight": wei2}})
     37                     print(G_Author.edges.data(), '\n')

KeyError: ('DP', 'TA')

主要错误来自这一行wei2 = wei[nod4, nod3] + 1，它是由 else 条件引起的。其中新数据为{ DP , TA }，由temp : [{'TA', 'AN'}, {'DP', 'TA'}] 证明。但是，当{'DP', 'TA'} 添加到G_Author 中时，顺序会切换并变为{'TA', 'DP'}。

我已经尝试将wei2 = wei[nod4, nod3] + 1 更改为wei2 = wei[nod3, nod4] + 1，但它只修复了第一个重复项。当另一个重复出现时，错误再次来自那 2 行代码。

也许有人可以帮我解决这个问题，或者有更好的解决方案来从重复的 set() 中生成权重？

【问题讨论】：

标签： python graph networkx edges weighted-graph

【解决方案1】：

您可能想要使用collections.Counter，它将计算迭代中特定键的出现次数：

from collections import Counter
from networkx import Graph

# create edge counts
t = [{'A', 'B'}, {'C', 'D'}, {'B', 'A'}]
# note that Counter does not use sets, so change data to tuples
c = Counter(map(tuple,t))
print(c)
# Counter({('A', 'B'): 2, ('C', 'D'): 1})

# add edges to the graph
G = Graph()
for (source, target), weight in c.items():
   G.add_edge(source, target, weight=weight)
print(G.edges(data=True))
# [('A', 'B', {'weight': 2}), ('C', 'D', {'weight': 1})]

【讨论】：