结合nltk.RegexpParser语法

时间:2017-10-08 23:26:19

标签: python parsing nlp nltk

作为我学习更多关于NLP的下一步,我尝试实现一种简单的启发式方法,可以提高结果,而不仅仅是简单的n-gram。

根据下面链接的斯坦福大学合着,他们提到通过一部分语音过滤器传递候选短语,这些过滤器只允许通过那些可能是“短语""与仅使用最常出现的二元组相比,它将产生更好的结果。 资料来源:搭配,第143-144页:https://nlp.stanford.edu/fsnlp/promo/colloc.pdf

第144页的表格有7种标记模式。按顺序,NLTK POS标签等效于:

JJ NN

NN

JJ JJ NN

JJ NN NN

NN JJ NN

NN NN NN

NN IN NN

在下面的代码中,当我独立应用下面的每个语法时,我可以得到所需的结果。然而,当我尝试组合相同的语法时,我没有收到所需的结果。

在我的代码中,您可以看到我取消注释一个句子,取消注释1个语法,运行它并检查结果。

我应该能够将所有句子组合起来,通过组合语法运行(在下面的代码中只有3个)并获得所需的结果。

  

我的问题是,如何正确组合语法?

我假设结合语法就像是' OR',找到这种模式,或者这种模式......

提前致谢。

import nltk

# The following sentences are correctly grouped with <JJ>*<NN>+. 
# Should see: 'linear function', 'regression coefficient', 'Gaussian random variable' and 
# 'cumulative distribution function'
SampleSentence = "In mathematics, the term linear function refers to two distinct, although related, notions"
#SampleSentence = "The regression coefficient is the slope of the line of the regression equation."
#SampleSentence = "In probability theory, Gaussian random variable is a very common continuous probability distribution."
#SampleSentence = "In probability theory and statistics, the cumulative distribution function (CDF) of a real-valued random variable X, or just distribution function of X, evaluated at x, is the probability that X will take a value less than or equal to x."

# The following sentences are correctly grouped with <NN.?>*<V.*>*<NN>
# Should see 'mean squared error' and # 'class probability function'. 
#SampleSentence = "In statistics, the mean squared error (MSE) of an estimator measures the average of the squares of the errors, that is, the difference between the estimator and what is estimated."
#SampleSentence = "The class probability function is interesting"

# The sentence below is correctly grouped with <NN.?>*<IN>*<NN.?>*. 
# should see 'degrees of freedom'.
#SampleSentence = "In statistics, the degrees of freedom is the number of values in the final calculation of a statistic that are free to vary."

SampleSentence = SampleSentence.lower()

print("\nFull sentence: ", SampleSentence, "\n")

tokens = nltk.word_tokenize(SampleSentence)
textTokens = nltk.Text(tokens)    

# Determine the POS tags.
POStagList = nltk.pos_tag(textTokens)    

# The following grammars work well *independently*
grammar = "NP: {<JJ>*<NN>+}"
#grammar = "NP: {<NN.?>*<V.*>*<NN>}"    
#grammar = "NP: {<NN.?>*<IN>*<NN.?>*}"


# Merge several grammars above into a single one below. 
# Note that all 3 correct grammars above are included below. 

'''
grammar = """
            NP: 
                {<JJ>*<NN>+}
                {<NN.?>*<V.*>*<NN>}
                {<NN.?>*<IN>*<NN.?>*}
        """
'''

cp = nltk.RegexpParser(grammar)

result = cp.parse(POStagList)

for subtree in result.subtrees(filter=lambda t: t.label() == 'NP'):
    print("NP Subtree:", subtree)    

1 个答案:

答案 0 :(得分:0)

如果我的评论是您正在寻找的,那么下面就是答案:

grammar = """
            NP: 
                {<JJ>*<NN.?>*<V.|IN>*<NN.?>*}"""