aspect_keywords.append(keywords_vector)
sentence = []
for s in sentences:
for i in s['text']:
if i in string.punctuation: # 如果字符是标点符号的话就将其替换为空格
s['text'] = s['text'].replace(i, " ")
sentence.append(s['text'])
sents = [word_tokenize(sent) for sent in sentence]
corpus = TextCollection(sents)
tf_idf = []
for sen in sents:
td = []
for data in sen:
elem = []
data = data.lower()
if data not in stop_words:
# print(data)
td.append(corpus.tf_idf(data, corpus))
tf_idf.append(td)
for aspect in aspect_keywords:
for vector in aspect[1:]:
print(
deal_data.cosine(aspect[0], vector) *
corpus.tf_idf('food', corpus))
print('\n')
sentences_vector = []
for w in sents[i]:
w = w.lower()
if w not in stop_words:
try:
word_index = list(words_index.keys())[list(
words_index.values()).index(w)]
sentences_vector.append(wordVectors[word_index])
except ValueError:
continue
count = []
for aspect in aspect_vector:
word_aspect_cosine = []
ci = 0
for j in range(len(sentences_vector)):
data_td = deal_data.cosine(aspect,
sentences_vector[j]) * tf_idf[i][j]
word_aspect_cosine.append(data_td)
if data_td > 0.0025:
ci = ci + 1
count.append(ci)
aspect_cosine.append(word_aspect_cosine)
count_s.append(count)
# # print(aspect_cosine)
# cosine.append(aspect_cosine)
print(len(sents))
print(len(sentences))
print(len(count_s))
#
count_a = 0
aspect_cosine = []
for w in s_w:
w = w.lower()
if w not in stop_words:
try:
word_index = list(words_index.keys())[list(
words_index.values()).index(w)]
sentences_vector.append(wordVectors[word_index])
except ValueError:
continue
count = [s['id']]
for aspect in aspect_vector:
word_aspect_cosine = []
i = 0
for word_vector in sentences_vector:
word_aspect_cosine.append(deal_data.cosine(aspect, word_vector))
if deal_data.cosine(aspect, word_vector) > 0.70:
i = i + 1
count.append(i)
aspect_cosine.append(word_aspect_cosine)
count_s.append(count)
# print(aspect_cosine)
# cosine.append(aspect_cosine)
#
for data in count_s:
print(data)
count_a = 0
for i in range(len(sentences)):
index = [
average_cosine = [s['id']]
for w in s_w:
w = w.lower()
if w not in stop_words:
try:
word_index = list(words_index.keys())[list(
words_index.values()).index(w)]
sentences_vector.append(wordVectors[word_index])
except ValueError:
continue
for data in aspect_keywords:
word_cosine = []
for word_vector in sentences_vector:
sum_cosine = 0
for i in range(len(data)):
sum_cosine = sum_cosine + deal_data.cosine(
data[i], word_vector)
word_cosine.append(sum_cosine / len(data))
if len(sentences_vector) != 0:
average_cosine.append(sum(word_cosine) / len(word_cosine))
cosine.append(average_cosine)
# print(cosine)
# 计算准确率accuracy:两个及以上aspect的句子怎么算？
# print(len(sentences))
# print(len(cosine))
# for i in range(len(sentences)):
# print(cosine[i])
# print(sentences[i])
w = w.lower()
if w not in stop_words:
try:
word_index = list(words_index.keys())[list(
words_index.values()).index(w)]
sentences_vector.append(wordVectors[word_index])
except ValueError:
continue
count = []
for data in aspect_keywords:
word_cosine = []
c = 0
for word_vector in sentences_vector:
label = set()
for i in range(len(data)):
if deal_data.cosine(data[i], word_vector) > 0.7:
label.add(1)
if 1 in label:
c = c + 1
count.append(c)
count_s.append(count)
for count in count_s:
print(count)
count_a = 0
for i in range(len(sentences)):
index = [j for j, data in enumerate(count_s[i]) if data == max(count_s[i])]
label_max = set()
for data in index:
s_w = s['text'].split()
sentences_vector = []
average_cosine = [s['id']]
for w in s_w:
w = w.lower()
if w not in stop_words:
try:
word_index = list(words_index.keys())[list(
words_index.values()).index(w)]
sentences_vector.append(wordVectors[word_index])
except ValueError:
continue
for aspect in aspect_vector:
sum_cosine = 0
for word_vector in sentences_vector:
sum_cosine = sum_cosine + deal_data.cosine(aspect, word_vector)
if len(sentences_vector) != 0:
average_cosine.append(sum_cosine / len(sentences_vector))
cosine.append(average_cosine)
# 计算准确率accuracy:两个及以上aspect的句子怎么算？
# print(len(sentences))
# print(len(cosine))
# for i in range(len(sentences)):
# print(cosine[i])
# print(sentences[i])
count = 0
count_empty = 0
for i in range(len(sentences)):