고급 Feacture 엔지니어링¶

1 Gensim 을 활용한 NLP 알고리즘¶

Word2Vec

! pip install gensim

In [9]:

from gensim.models import Word2Vec
w2v_model = Word2Vec.load('./data/model.bin')
w2v_model.wv.most_similar(positive=['woman','king'], negative=['man'])

Out[9]:

[('chryses', 0.6371129751205444),
 ('priest', 0.6282371282577515),
 ('nymph', 0.6165897250175476),
 ('thanks', 0.6120550632476807),
 ('dishonored', 0.6062030792236328),
 ('narrate', 0.605045735836029),
 ('angered', 0.6038438677787781),
 ('chieftains', 0.6015218496322632),
 ('appease', 0.6003137826919556),
 ('akhilleus', 0.6002722978591919)]

In [10]:

w2v_model.wv.similarity('woman','man')

Out[10]:

0.3998656

In [22]:

# sorted(w2v_model.wv.vocab.keys(), reverse=False)[:14]
len(w2v_model.wv.vocab.keys())

Out[22]:

2 Gensim 을 활용한 유사도 분석실습¶

Word2Vec

! pip install gensim
"Ice" 와 "Fire" 의 문학내 유사도 측정

In [25]:

w2v_model.wv.most_similar('stark')

Out[25]:

[('principals', 0.9933313131332397),
 ('threatening', 0.9859204292297363),
 ('distorting', 0.9620187878608704),
 ('freeport', 0.9174789190292358),
 ('stood', 0.8492887616157532),
 ('extend', 0.8167773485183716),
 ('douglas', 0.767949104309082),
 ('1858', 0.7635020017623901),
 ('conspiracy', 0.762062668800354),
 ('values', 0.7606658339500427)]