/
search_engine.py
executable file
·221 lines (177 loc) · 6.8 KB
/
search_engine.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
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
#! /usr/bin/python
#
# Usage:
# python search_engine.py seed_URL n_pages [optional fraction of terms to use]
#
# Example:
# python search_engine.py http://en.wikipedia.org/wiki/Business_intelligence 100 .95
import sys
import random
import crawler
import parser
import indexing
import TFIDF
import jaccard
#############################################################
seed = sys.argv[1]
n_pages = int(sys.argv[2])
term_fraction = None if len(sys.argv) < 4 else float(sys.argv[3])
print 'Downloading and indexing documents'
direct_index = indexing.get_index (crawler.recursive_download (seed, n_pages))
print len(direct_index), ' documents downloaded.'
print 'Now computing the inverted index'
inverted_index = indexing.invert_index (direct_index)
print len(inverted_index), 'terms indexed.'
# For convenience, remember the number of terms and documents
n_documents = len(direct_index)
n_terms = max(t['termid'] for t in inverted_index.values()) + 1
# Compute the optional IDF threshold
idf_threshold = None if not term_fraction else indexing.IDF_threshold (inverted_index, term_fraction)
if idf_threshold:
print "IDF threshold set at", idf_threshold
#############################################################
# Display the contents of the index, term by term
'''
for t, l in inverted_index.items():
# Write it
sys.stdout.write ('Term: %s, ID: %d, IDF: %.5f\n'
% (t, l['termid'], l['idf']))
# Dor all documents appearing in the inverted index
for d, c, tf in l['occurrences']:
# write the document triplet
sys.stdout.write ('\tDoc ID: %03d, # occurrences: %5d, TF: %.8f\n'
% (d, c, tf))
exit()
'''
##############################################################
print 'Computing TFIDF representations of documents in the corpus'
TFIDFs = TFIDF.compute_all_TFIDFs (inverted_index, idf_threshold)
# Print all cosine similarities between documents
'''
similarities = [[TFIDF.cosine_similarity(d1,d2) for d2 in TFIDFs] for d1 in TFIDFs]
print similarities
exit()
'''
# Given a query, compute its TFIDF representation
print 'Computing query\'s TFIDF representation'
query = 'business became meaningful'
query_terms = parser.extract_terms (query)
q_TFIDF = TFIDF.compute_new_TFIDF (query_terms, inverted_index, idf_threshold)
# Warn if the query is empty due to a high IDF threshold
if len(q_TFIDF) == 0:
print '*** WARNING *** Empty query, IDf threshold too high'
'''
print q_TFIDF
exit()
'''
######################################################################
print 'Computing set representations of documents in the corpus'
sets = jaccard.compute_all_sets (inverted_index, idf_threshold)
# Print all jaccard similarities between documents
similarities = [[jaccard.jaccard_coefficient(d1,d2) for d2 in sets] for d1 in sets]
'''
print similarities
exit()
'''
# Compute the query's set representation
print 'Computing query\'s set representation'
q_set = jaccard.compute_new_set (query_terms, inverted_index, idf_threshold)
'''
print q_set
exit()
'''
print 'Computing document ranking based on increasing Euclidean distance from query in TFIDF space'
euclidean_ranking = [(i, TFIDF.Euclidean_distance (q_TFIDF, TFIDFs[i])) for i in range(n_documents)]
euclidean_ranking.sort (key = lambda t: t[1])
print 'Computing document ranking based on decreasing cosine similarity with query in TDIDF space'
cosine_ranking = [(i, TFIDF.cosine_similarity (q_TFIDF, TFIDFs[i])) for i in range(n_documents)]
cosine_ranking.sort (key = lambda t: t[1], reverse = True)
print 'Computing document ranking based on decreasing Jaccard coefficient with query'
jaccard_ranking = [(i, jaccard.jaccard_coefficient (q_set, sets[i])) for i in range(n_documents)]
jaccard_ranking.sort (key = lambda t: t[1], reverse = True)
print 'Resulting top rankings:\nEuclidean\t\tCosine\t\t\tJaccard'
for i in range(10):
print '%d\t%f\t%d\t%f\t%d\t%f' % (
euclidean_ranking[i][0],
euclidean_ranking[i][1],
cosine_ranking[i][0],
cosine_ranking[i][1],
jaccard_ranking[i][0],
jaccard_ranking[i][1])
'''
#calculate inverted index of given ranking
def invert_ranking (rk):
return [[docid for docid,key in rk].index(i)
for i in range(n_documents)]
euclidean_inverse_ranking = invert_ranking (euclidean_ranking)
cosine_inverse_ranking = invert_ranking (cosine_ranking)
jaccard_inverse_ranking = invert_ranking (jaccard_ranking)
#print in a file caluculated inverted index
f = open ('ranking.txt','w')
for docid in range(n_documents):
f.write ('%d\t%d\t%d\t%d\n' % (
docid,
euclidean_inverse_ranking[docid],
cosine_inverse_ranking[docid],
jaccard_inverse_ranking[docid]
))
f.close()
'''
#estimate jaccard coeficients with permutation and probability
n_perm = 20
P = [range(n_terms) for i in range(n_perm)]
#create n_perm random permutation
for p in P:
random.shuffle(p)
#create a list of all min for every doc
minima = [
[min (p[termid] for termid in s) for p in P]
for s in sets
]
#count the equality of permutation for 2 files
def count_equalities (a, b):
c=0
for i in range(len(a)):
if a[i] == b[i]:
c += 1
return c
'''
#estimate jaccard with probability by count equalitisi in minma divided by total attempt
#for every pair of docs
jaccard_estimate = [
[
float(count_equalities(m1, m2))/n_perm for m2 in minima
]
for m1 in minima
]
#compare jaccard_estimate with jaccard_similarity in a file
#you con plot with gnuplot to understand
f = open("jaccard.txt", "w")
for i in range(n_documents):
for j in range(n_documents):
f.write('%f\t%f\n' % (similarities[i][j], jaccard_estimate[i][j]))
f.close()
'''
#for all perm i take the min for the query set (q_sset)
q_minima = [min (p[termid] for termid in q_set) for p in P]
approximate_jaccard_ranking = [[i, float(count_equalities(q_minima, minima[i]))/n_perm] for i in range(n_documents)]
jaccard_ranking.sort(key = lambda t: t[1], reverse = True)
#calculate inverted index of given ranking
def invert_ranking (rk):
return [[docid for docid,key in rk].index(i)
for i in range(len(rk))]
euclidean_inverse_ranking = invert_ranking (euclidean_ranking)
cosine_inverse_ranking = invert_ranking (cosine_ranking)
jaccard_inverse_ranking = invert_ranking (jaccard_ranking)
approximate_jaccard_inverse_ranking = invert_ranking (approximate_jaccard_ranking)
#print in a file caluculated inverted index
f = open ('ranking.txt','w')
for docid in range(n_documents):
f.write ('%d\t%d\t%d\t%d\t%d\n' % (
docid,
euclidean_inverse_ranking[docid],
cosine_inverse_ranking[docid],
jaccard_inverse_ranking[docid],
approximate_jaccard_inverse_ranking[docid]
))
f.close()