5. word2vecによる類似単語 - hpscript/nlp GitHub Wiki

gensimインストール

$ pip3 install gensim==3.8.1

word2vecのファイルをDL

$ wget http://public.shiroyagi.s3.amazonaws.com/latest-ja-word2vec-gensim-model.zip $ unzip latest-ja-word2vec-gensim-model.zip

app.py

# -*- coding: utf-8 -*-
from gensim.models import word2vec

model = word2vec.Word2Vec.load('word2vec.gensim.model')
results = model.wv.most_similar(positive=['日本'])
for result in results:
	print(result)

$ python3 app.py ('韓国', 0.7088127732276917) ('台湾', 0.6461570262908936) ('日本国内', 0.6403890252113342) ('欧米', 0.6350583434104919) ('日本国外', 0.6200590133666992) ('台湾出身', 0.6174061894416809)