依赖树基本上是一个图,所以如果你想找到 ROOT 的(最短)路径,你需要使用一些基于图的库,比如networkx
. 假设您要提取从令牌“望远镜”到根的路径。然后你可以尝试做这样的事情:
import spacy
import networkx
sentence = "I saw the man with a telescop"
nlp = spacy.load('en_core_web_sm')
doc = nlp(sentence)
edges = []
for sent in doc.sents:
for token in sent:
print("{}\t{}\t{}\t{}".format(token.i, token.text, token.head, token.dep_))
if token.dep_ == "ROOT":
target = token.text
for child in token.children:
edges.append(("{0}".format(token.lower_), "{0}".format(child.lower_)))
graph = networkx.Graph(edges)
print(nx.shortest_path(graph, source="telescop", target=target))
结果:
0 I saw nsubj
1 saw saw ROOT
2 the man det
3 man saw dobj
4 with saw prep
5 a telescop det
6 telescop with pobj
['telescop', 'with', 'saw']