eval_w2v_avg.py

from gensim.similarity_learning import WikiQAExtractor

wikiqa = WikiQAExtractor(os.path.join("..", "data", "WikiQACorpus", "WikiQA-train.tsv"))
data = wikiqa.get_data()

# Below commented code is for making a dict for word vectors and pickling it
# w2v = {}

# with open('glove.6B.50d.txt') as f:
# 	for line in f:
# 		string_array = np.array(line.split()[1:])
# 		string_array = [float(i) for i in string_array]
# 		w2v[line.split()[0]] = string_array

# with open('w2v.pkl', 'wb') as f:
# 	pickle.dump(w2v, f)

with open('w2v.pkl', 'rb') as f:
	w2v = pickle.load(f)

def sent2vec(sentence):
	vec_sum = []

	for word in sentence.split():
		if word in w2v:
			vec_sum.append(w2v[word])

	return np.mean(np.array(vec_sum), axis=0)

def cos_sim(vec1, vec2):
	return np.sum(vec1*vec2)/(np.linalg.norm(vec1)*np.linalg.norm(vec2))

big_y_true = []
big_y_pred = []

for doc in data:
	y_true = []
	y_pred = []

	for query, doc, label in doc:
		y_pred.append(cos_sim(sent2vec(query), sent2vec(doc)))
		y_true.append(label)

	big_y_true.append(y_true)
	big_y_pred.append(y_pred)

big_y_pred = np.array(big_y_pred)
big_y_true = np.array(big_y_true)


n_correct = 0

for y_pred, y_true in zip(big_y_pred, big_y_true):
	if (np.argmax(y_true) == np.argmax(y_pred)):
		n_correct += 1


print("Accuracy : ", n_correct/len(big_y_true))
# Accuracy :  0.2867264997638167