Take the Brown corpus in NLTK and choose the government subcorpus:
from nltk.corpus import brown
nltk.download('brown')
gov_text = brown.words(categories='government')Preporcess the text by filtering words shorter than three letters and words that are marked as stopwords.
Find ten best trigrams according to the log-likelihood association score.
The output should be a list of the tuples (trigrams) you have found, with each trigram on a new line. For example,
[('First word', 'Second word', 'Third word'),
('First word', 'Second word', 'Third word'),
('First word', 'Second word', 'Third word')]