w2v-training.py

import pandas as pd
from gensim.models import Word2Vec
import logging
import re

# Diplay log
logging.basicConfig(format='%(asctime)s : %(levelname)s : %(message)s', level=logging.INFO)

# Load data
data = pd.read_csv('training-data.csv', header=None)

# Preprocess text
def tokenize(sentence):
    remove_dots = re.sub("[.]", "", sentence.lower())
    return re.findall("[A-Za-z]{2,}", remove_dots)

sentences = []
for sentence in data[0].values:
    sentences.append(tokenize(sentence))

# Train and save word2vec model
model = Word2Vec(sentences, min_count=5, size=100, iter=500, seed=96)
model.save('word2vec.bin')

# Load and test word2vec model
loaded_w2v = Word2Vec.load('word2vec.bin')
print(loaded_w2v.wv.most_similar(positive=['pilot']))
print(loaded_w2v.wv['pilot'])