Python实现文章自动生成

Python实现文章自动生成

---

  下面的Python程序实现了通过从网页抓取一篇文章，然后根据这篇文章来生成新的文章，这其中的原理就是基于概率统计的文本分析。
  过程大概就是网页抓取数据->统计分析->生成新文章。网页抓取数据是通过BeautifulSoup库来抓取网页上的文本内容。统计分析这个首先需要使用ngram模型来把文章进行分词并统计频率。因为文章生成主要依据马尔可夫模型，所以使用了2-gram，这样可以统计出一个单词出现在另一个单词后的概率。生成新文章是基于分析大量随机事件的马尔可夫模型。随机事件的特点是在一个离散事件发生之后，另一个离散事件将在前一个事件的条件下以一定的概率发生。

from urllib.request import urlopen
from random import randint
from bs4 import BeautifulSoup
import re

def wordListSum(wordList):
    sum = 0
    for word, value in wordList.items():
        sum = sum + value
    return sum


def retrieveRandomWord(wordList):

    randomIndex = randint(1, wordListSum(wordList))
    for word, value in wordList.items():
        randomIndex -= value
        if randomIndex <= 0:
            return word



def buildWordDict(text):
    text = re.sub('(\n|\r|\t)+', " ", text)
    text = re.sub('\"', "", text)

    punctuation = [',', '.', ';', ':']
    for symbol in punctuation:
        text = text.replace(symbol, " " + symbol + " ")

    words = text.split(' ')

    words = [word for word in words if word != ""]
    wordDict = {}
    for i in range(1, len(words)):
        if words[i-1] not in wordDict:
            wordDict[words[i-1]] = {}
        if words[i] not in wordDict[words[i-1]]:
            wordDict[words[i-1]][words[i]] = 0
        wordDict[words[i-1]][words[i]] = wordDict[words[i-1]][words[i]] + 1

    return wordDict

def randomFirstWord(wordDict):
    randomIndex = randint(0, len(wordDict))
    return list(wordDict.keys())[randomIndex]

html = urlopen("http://www.guancha.cn/america/2017_01_21_390488_s.shtml")
bsObj = BeautifulSoup(html, "lxml")
ps = bsObj.find("div", {"id": "cmtdiv3523349"}).find_next_siblings("p");
content = ""
for p in ps:
    content = content + p.get_text()
text = bytes(content, "UTF-8")
text = text.decode("ascii", "ignore")
wordDict = buildWordDict(text)

length = 100
chain = ""
currentWord = randomFirstWord(wordDict)
for i in range(0, length):
    chain += currentWord + " "
    currentWord = retrieveRandomWord(wordDict[currentWord])

print(chain)

buildWordDict(text)函数接收文本内容，生成的内容如下

{‘itself’: {‘,’: 1}, ‘night’: {‘sky’: 1}, ‘You’: {‘came’: 1, ‘will’: 1}, ‘railways’: {‘all’: 1}, ‘government’: {‘while’: 1, ‘,’: 1, ‘is’: 1}, ‘you’: {‘now’: 1, ‘open’: 1, ‘down’: 1, ‘with’: 1, ‘.’: 6, ‘,’: 1, ‘that’: 1},

主要就是生成一个字典，键是文章中所有出现的词语，值其实也是一个字典，这个字典是所有直接出现在键后边的词语及其出现的频率。这个函数就是ngram模型思想的运用。
retrieveRandomWord(wordList)函数的wordList代表的是出现在上一个词语后的词语列表及其频率组成的字典，然后根据统计的概率随机生成一个词。这个函数是马尔可夫模型的思想运用。

然后运行这个程序会生成一个长度为100的文章，如下面所示

fail . We will stir ourselves , but we will never before . Do not share one heart and pleasant it back our jobs . We are infused with the orderly and railways all of the gangs and robbed our jobs for their success will determine the civilized world . We will their success will be a great men and highways and millions to all bleed the world . It belongs to great national effort to defend our products , constantly complaining , D . We will be ignored again . It belongs to harness the expense of America .

生成的文章看起来语法混乱，这也难怪，因为只是抓取分析统计了一篇的文章。我想如果可以抓取足够多的英文文章，数据集足够大那么语法准确度会大大提高。