networkx - 将单词列表转换为最小生成树算法的问题图

Question

我一直在尝试从单词列表或单词字符串中构造一个问题加权图，以将单词字符串指定给命名元组列表，其中每个可能的单词组合为头和尾，附加头“根”

输入：

['john', 'saw', 'mary', 'root']

所需输出：

[Arc('root',weight,'saw'),Arc('root',weight,'john'),Arc('root', weight,'mary'),
 Arc('saw', weight,'john'), Arc('john', weight, 'saw'), Arc('saw', weight,'mary'),Arc('john', weight,'mary'),
 Arc('mary', weight,'john'),Arc('saw', weight,'mary'),Arc('mary', weight,'saw')]

代码：

from collections import defaultdict, namedtuple 

Arc = namedtuple('Arc', ('head', 'weight', 'tail'))

def Constrain_graph(sentence):
    Arc = namedtuple('Arc', ('head', 'weight', 'tail'))
    C_graph=[]
    for wordindex in range(1, len(sentence)):
        G = nx.DiGraph()
        G.nodes(sentence[wordindex])

        #G.add_node(nod)
        G.add_nodes_from(range(1,len(sentence)))
        C_graph=nx.Graph()
    return C_graph

我需要输出像上面提到的那样