使用python将句子分成段落

我需要使用python将视频标题的句子分成段落。我尝试使用nltk.tokenize.textiling，但没有得到任何结果。 以下是全文摘录：

– [Voiceover] Bob Dylan is,
you must be 20 years old now,
aren't you?
– [Voiceover] Yeah, I must be 20.
(laughing)
– [Voiceover] Are you?
– [Voiceover] Yeah, I'm 20, I'm 20.
(guitar music)
My hands are cold.
It's a pretty cold studio.
– [Voiceover] The coldest studio.
– [Voiceover] Usually can do this.
There I just want to do it once.
(guitar strumming)
– [Voiceover] When I first heard Bob Dylan
was, I think, about three
years ago in Minneapolis.
– [Voiceover] At that time I
was just sort of doing nothing.
I was there working, I guess.
I was making pretend I was
going to school out there.
I'd just come there from South Dakota.
– [Voiceover] You've sung
now at Goody's here in town.
Have you sung at any of the coffee houses?

---

看起来很简单，你可以用正则表达式， 我不知道你想要哪种格式，但这里有一个例子

import re

sentence = """
– [Voiceover] Bob Dylan is,
you must be 20 years old now,
aren't you?
– [Voiceover] Yeah, I must be 20.
(laughing)
– [Voiceover] Are you?
– [Voiceover] Yeah, I'm 20, I'm 20.
(guitar music)
My hands are cold.
It's a pretty cold studio.
– [Voiceover] The coldest studio.
– [Voiceover] Usually can do this.
There I just want to do it once.
(guitar strumming)
– [Voiceover] When I first heard Bob Dylan
was, I think, about three
years ago in Minneapolis.
– [Voiceover] At that time I
was just sort of doing nothing.
I was there working, I guess.
I was making pretend I was
going to school out there.
I'd just come there from South Dakota.
– [Voiceover] You've sung
now at Goody's here in town.
Have you sung at any of the coffee houses?

"""

start_re = re.compile(r'\–\s\[.*?\]')
result = re.split(start_re,sentence)
result = filter(lambda x:x, [s.replace('\n','').strip() for s in result])
print result

输出

["Bob Dylan is,you must be 20 years old now,aren't you?", 'Yeah, I must be 20.(laughing)', 'Are you?', "Yeah, I'm 20, I'm 20.(guitar music)My hands are cold.It's a pretty cold studio.", 'The coldest studio.', 'Usually can do this.There I just want to do it once.(guitar strumming)', 'When I first heard Bob Dylanwas, I think, about threeyears ago in Minneapolis.', "At that time Iwas just sort of doing nothing.I was there working, I guess.I was making pretend I wasgoing to school out there.I'd just come there from South Dakota.", "You've sungnow at Goody's here in town.Have you sung at any of the coffee houses?"]

---

我从来没有想到过这样的事。在任何情况下，你都在试图解析人类对话的文本吗？将这些段落转换成段落似乎完全是无稽之谈——段落是组织人类写作的一种方式，不适用于人类对话。你如何将句子分割成段落？段落是由句子组成的，但句子是由单词组成的。可以使用NLTK模块找到句子，如果句子之间的主题完全改变，那么它就是新段落的指示。这就是文本拼接的工作原理。我认为这是为了找出句子，而不是段落。我知道段落识别是一项艰巨的任务，边界也很微妙，但如果您能向我推荐任何其他库，而不是nltk.tokenize.textiling，我将不胜感激。