80 lines
2.5 KiB
Python
80 lines
2.5 KiB
Python
from tqdm import tqdm
|
|
import regex as re
|
|
from nltk.tokenize import word_tokenize
|
|
from english_words import get_english_words_set
|
|
import kenlm
|
|
from math import log10
|
|
import pickle
|
|
|
|
path = 'kenlm_model.binary'
|
|
model = kenlm.Model(path)
|
|
|
|
with open('V.pickle', 'rb') as handle:
|
|
V_counter = pickle.load(handle)
|
|
|
|
def clean_string(text):
|
|
text = text.lower()
|
|
text = re.sub(r" -\\*\\n", "", text)
|
|
text = re.sub(r"\\n", " ", text)
|
|
text = text.strip()
|
|
return text
|
|
|
|
|
|
def predict_probs(w1, w2, w4, w5):
|
|
best_scores = []
|
|
pred_str = ""
|
|
# for word in get_english_words_set(['web2'], lower=True):
|
|
for word in V_counter:
|
|
text = ' '.join([w1, w2, word, w4, w5])
|
|
text_score = model.score(text, bos=False, eos=False)
|
|
if len(best_scores) < 5:
|
|
best_scores.append((word, text_score))
|
|
else:
|
|
worst_score = best_scores[-1]
|
|
if worst_score[1] < text_score:
|
|
best_scores[-1] = (word, text_score)
|
|
best_scores = sorted(best_scores, key=lambda tup: tup[1], reverse=True)
|
|
|
|
for word, prob in best_scores:
|
|
pred_str += f'{word}:{prob} '
|
|
pred_str += f':{log10(0.99)}'
|
|
return pred_str
|
|
|
|
def get_word_predictions(w1, w2,):
|
|
for word in get_english_words_set(['web2'], lower=True):
|
|
sentence = w1 + ' ' + word + ' ' + w2
|
|
text_score = model.score(sentence, bos=False, eos=False)
|
|
yield((word, text_score))
|
|
|
|
def argmax(w1,w2):
|
|
# get top 10 predictions from predict_line
|
|
top_10 = sorted(list(get_word_predictions(w1,w2)), key=lambda x: -x[1])[:4]
|
|
output_line = " ".join(["{}:{:.8f}".format(w, p) for w, p in top_10])
|
|
return output_line
|
|
|
|
def run_predictions(source_folder):
|
|
print(f"Run predictions on {source_folder} data...")
|
|
|
|
with open(f"{source_folder}/in.tsv", encoding="utf8", mode="rt") as file:
|
|
train_data = file.readlines()
|
|
|
|
with open(f"{source_folder}/out.tsv", "w", encoding="utf-8") as output_file:
|
|
for line in tqdm(train_data):
|
|
line = line.split("\t")
|
|
|
|
l1 = clean_string(line[-2])
|
|
l2 = clean_string(line[-1])
|
|
|
|
if not l1 or not l2:
|
|
out_line = "the:0.2 be:0.2 to:0.2 of:0.1 and:0.1 a:0.1 :0.1"
|
|
else:
|
|
w1, w2 = word_tokenize(l1)[-2:]
|
|
w3, w4 = word_tokenize(l2)[:2]
|
|
out_line = predict_probs(w1, w2, w3, w4)
|
|
|
|
output_file.write(out_line + "\n")
|
|
|
|
|
|
run_predictions("dev-0")
|
|
run_predictions("test-A")
|