retrieval.py

import sys
import gensim
import numpy as np

W2V_PATH = sys.argv[1]

def avg_sentence(sentence, wv):
    v = np.zeros(300)
    for w in sentence:
        if w in wv:
            v += wv[w]
    return v / len(sentence)

def cosine_sim(a, b):
    return np.dot(a, b) / (np.linalg.norm(a) * np.linalg.norm(b))

url_descr = [
    ('http://tny.im/a13', 'we don\'t care'),
    ('https://tny.im/a0-', 'i love you'),
    ('https://tny.im/a12', 'relax take it easy'),
    ('https://tny.im/a16', 'that is embarrassing'),
    ('https://tny.im/a10', 'screw you guys')
]

model = gensim.models.KeyedVectors.load_word2vec_format(W2V_PATH, binary=True)

inputv = avg_sentence(input().split(), model.wv)
avgs = list(map(lambda p: p + (avg_sentence(p[1].split(), model.wv),), url_descr))
sims = list(map(lambda p: p[:2] + (cosine_sim(inputv, p[2]),), avgs))

most_similar_meme = sorted(sims, key=lambda p: p[2], reverse=True)
print(most_similar_meme)

retrieval_io

Input: i like you
Output: [
  ('https://tny.im/a0-', 'i love you', 0.87972317862080773), 
  ('https://tny.im/a10', 'screw you guys', 0.66992365073780347), 
  ('http://tny.im/a13', "we don't care", 0.56639891559620026), 
  ('https://tny.im/a12', 'relax take it easy', 0.40517121688823338), 
  ('https://tny.im/a16', 'that is embarrassing', 0.2843743794717129)
]