-
Notifications
You must be signed in to change notification settings - Fork 0
/
frontend_utils.py
185 lines (142 loc) · 6.1 KB
/
frontend_utils.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
import re
import math
from collections import Counter
from nltk.stem.porter import *
from nltk.corpus import stopwords
from nltk.corpus import wordnet
def query_expansion(query):
synonyms = {}
hyponyms = {}
for word in query:
word_synsets = wordnet.synsets(word)
for synset in word_synsets:
syn = synset.lemmas()[0].name().lower()
if syn and syn != word:
synonyms[word] = syn
break
for word in query:
word_synsets = wordnet.synsets(word)
for synset in word_synsets:
try:
hyp = synset.hyponyms()[0].name()
word_synset = wordnet.synset(hyp)
hypernyms = word_synset.hypernyms()
lst = ([synset.name().split(".")[0] for synset in hypernyms])
h = lst[0].lower()
if h != word:
hyponyms[word] = h
except:
pass
for _, val in synonyms.items():
if val not in query:
query.append(val)
for _, val in hyponyms.items():
if val not in query:
query.append(val)
return query
def tokenize(text, STEMMING=False, QUERYEXP=False):
RE_WORD = re.compile(r"""[\#\@\w](['\-]?[\w,]?[\w.]?(?:['\-]?[\w,]?[\w])){0,24}""", re.UNICODE)
english_stopwords = frozenset(stopwords.words('english'))
corpus_stopwords = ["category", "references", "also", "external", "links",
"may", "first", "see", "history", "people", "one", "two",
"part", "thumb", "including", "second", "following",
"many", "however", "would", "became"]
all_stopwords = english_stopwords.union(corpus_stopwords)
tokens = [token.group() for token in RE_WORD.finditer(text.lower())]
if QUERYEXP:
tokens = query_expansion(tokens)
if STEMMING:
stemmer = PorterStemmer()
list_of_tokens = [stemmer.stem(x) for x in tokens if x not in all_stopwords]
else:
list_of_tokens = [x for x in tokens if x not in all_stopwords]
return list_of_tokens
def old_tokenize(text):
RE_WORD = re.compile(r"""[\#\@\w](['\-]?\w){2,24}""", re.UNICODE)
english_stopwords = frozenset(stopwords.words('english'))
corpus_stopwords = ["category", "references", "also", "external", "links",
"may", "first", "see", "history", "people", "one", "two",
"part", "thumb", "including", "second", "following",
"many", "however", "would", "became"]
all_stopwords = english_stopwords.union(corpus_stopwords)
tokens = [token.group() for token in RE_WORD.finditer(text.lower())]
list_of_tokens = [x for x in tokens if x not in all_stopwords]
return list_of_tokens
def BM25(tokens, K, B, AVGDL, inverted_index, index_folder_url, DL, DL_LEN):
doc_BM25_value = Counter()
for token in tokens:
# calc idf for specific token
try:
token_df = inverted_index.df[token]
except:
continue
token_idf = math.log(DL_LEN/token_df,10)
# loading posting list with (word, (doc_id, tf))
posting_list = inverted_index.read_posting_list(token, index_folder_url)
for page_id, word_freq in posting_list:
#normalized tf (by the length of document)
try:
numerator = word_freq*(K+1)
denominator = word_freq + K*(1-B + (B*DL[page_id])/AVGDL)
doc_BM25_value[page_id] += token_idf*(numerator/denominator)
except:
pass
sorted_doc_BM25_value = doc_BM25_value.most_common()
return sorted_doc_BM25_value
def cossim(tokens, inverted_index, index_folder_url, DL, DL_LEN, NF):
# get frequency of each token in query
query_freq = Counter(tokens)
numerator = Counter()
query_denominator = 0
weight_token_query = 0
query_len = len(tokens)
for token in tokens:
# calc idf for specific token
try:
token_df = inverted_index.df[token]
except:
continue
token_idf = math.log(DL_LEN/token_df, 10)
# calc query_token_tf
tf_of_query_token = query_freq[token]/query_len
weight_token_query = tf_of_query_token*token_idf
query_denominator += math.pow(weight_token_query ,2)
# loading posting list with (word, (doc_id, tf))
posting_list = inverted_index.read_posting_list(token, index_folder_url)
for page_id, word_freq in posting_list:
#normalized tf (by the length of document)
try:
tf = (word_freq/DL[page_id])
weight_word_page = tf*token_idf
numerator[page_id] += weight_word_page*weight_token_query
except:
pass
cosim = Counter()
for page_id in numerator.keys():
cosim[page_id] = numerator[page_id]/((math.sqrt(query_denominator)*NF[page_id]))
sorted_doc_cossim_value = cosim.most_common()
return sorted_doc_cossim_value
def get_binary_score(tokens, inverted_index, index_folder_url):
# loading posting list with (word, (doc_id, tf))
posting_lists = inverted_index.get_posting_lists(tokens, index_folder_url)
tf_dict = {}
for posting in posting_lists:
for doc_id, _ in posting:
if doc_id in tf_dict:
tf_dict[doc_id] += 1
else:
tf_dict[doc_id] = 1
list_of_docs = sorted([(doc_id, score) for doc_id, score in tf_dict.items()], key=lambda x: x[1], reverse=True)
return list_of_docs
def get_power_score(tokens, inverted_index, index_folder_url):
# loading posting list with (word, (doc_id, tf))
posting_lists = inverted_index.get_posting_lists(tokens, index_folder_url)
tf_dict = {}
for posting in posting_lists:
for doc_id, tf in posting:
if doc_id in tf_dict:
tf_dict[doc_id] += tf
else:
tf_dict[doc_id] = tf
list_of_docs = sorted([(doc_id, score) for doc_id, score in tf_dict.items()], key=lambda x: x[1], reverse=True)
return list_of_docs