如何使用NLTK合并词性标注后的连续专有名词？

Got it, let's fix that issue where you're only extracting single proper nouns instead of merging consecutive ones like "Ronald McDonald" or "Central Park". Here's a simple, effective approach tailored to your use case:

Step-by-Step Solution #

First, let's start with your existing code to get the tagged tokens, then add logic to group consecutive NNP tags into full phrases:

import nltk
# Download required resources if you haven't already
nltk.download('punkt')
nltk.download('averaged_perceptron_tagger')

sentence = 'Ronald McDonald exercised at the Central Park on Monday.'
tokens = nltk.word_tokenize(sentence)
tagged = nltk.pos_tag(tokens)

# Initialize lists to track and store merged proper nouns
merged_proper_nouns = []
current_consecutive_nnp = []

# Iterate through each tagged word pair
for word, tag in tagged:
    # If the word is a proper noun, add it to our temporary list
    if tag == 'NNP':
        current_consecutive_nnp.append(word)
    else:
        # If we were collecting proper nouns, merge and save them
        if current_consecutive_nnp:
            merged_phrase = ' '.join(current_consecutive_nnp)
            merged_proper_nouns.append(merged_phrase)
            current_consecutive_nnp = []  # Reset for next group

# Don't forget to catch any remaining proper nouns at the end of the sentence
if current_consecutive_nnp:
    merged_proper_nouns.append(' '.join(current_consecutive_nnp))

print(merged_proper_nouns)
# Output: ['Ronald McDonald', 'Central Park', 'Monday']

How This Works #

• Temporary Tracking: We use current_consecutive_nnp to hold words that are consecutive proper nouns as we iterate through the tagged list.
• Merge & Save: When we hit a non-proper noun, we check if we've been collecting NNPs. If so, we join them into a single phrase and add it to our final list, then reset the temporary list.
• Final Check: After the loop ends, we make sure to add any remaining NNPs (in case the sentence ends with a proper noun, like "Monday" in your example).

Bonus: Handle Plural Proper Nouns #

If you also want to include plural proper nouns tagged as NNPS, just adjust the condition to check if the tag starts with "NNP":

if tag.startswith('NNP'):
    current_consecutive_nnp.append(word)

This will catch both singular (NNP) and plural (NNPS) proper nouns, so phrases like "The Smiths" will be merged correctly too.

内容的提问来源于stack exchange，提问作者user9092346