forked from mingzi151/SummPip
/
utils.py
178 lines (156 loc) · 5.96 KB
/
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
# -*- coding: utf-8 -*-
"""
Created on 2020-01-16 10:17 PM
author : michelle
"""
import numpy as np
import os
import takahe
import nltk.data
import spacy
sent_detector = nltk.data.load('tokenizers/punkt/english.pickle')
spacynlp=spacy.load("en_core_web_sm")
def rank_sent_Lexrank(lxr,sentences_list,cutoff=1.5):
scores = list(lxr.rank_sentences(sentences_list,threshold=None,fast_power_method=True))
# idx_to_slice=[sscores.index(s) for s in scores if s> cutoff]
idx_to_slice=[ i for i, score in enumerate(scores) if score>cutoff]
sentences_list=[sentences_list[idx] for idx in idx_to_slice]
# print("ranked sentence:", sentences_list)
return sentences_list
def rank_sent_centroid(centroid_rank,sentences_list):
sentences_list=centroid_rank.centroidRank(sentences_list)
return sentences_list
# get word embeddings
def get_w2v_embeddings(filename):
word_embeddings = {}
f = open(filename, encoding='utf-8')
for line in f:
values = line.split()
word = values[0]
coefs = np.asarray(values[1:], dtype='float32')
word_embeddings[word] = coefs
f.close()
return word_embeddings
def get_sentence_embedding(sent, word_embeddings):
sent=sent.lower()
eps=1e-10
#print(sent)
if len(sent) != 0:
vectors = [word_embeddings.get(w, np.zeros((100,))) for w in sent.split()]
v=np.mean(vectors, axis=0)
else:
v = np.zeros((100,))
v = v + eps
return v
def get_SentNode_embedding(sentences_list, word_embeddings):
emb_sentence_vectors=np.zeros([len(sentences_list),100])
for count, sent in enumerate(sentences_list):
emb_sen=get_sentence_embedding(sent, word_embeddings)
emb_sentence_vectors[count,]=emb_sen
return emb_sentence_vectors
def build_similarity_matrix(emb_sentence_vectors):
from sklearn.metrics.pairwise import cosine_similarity
sim_mat = np.zeros([len(emb_sentence_vectors), len(emb_sentence_vectors)])
for i in range(len(emb_sentence_vectors)):
for j in range(len(emb_sentence_vectors)):
if i != j:
sim_mat[i][j] = cosine_similarity(emb_sentence_vectors[i].reshape(1,100), emb_sentence_vectors[j].reshape(1,100))[0,0]
return sim_mat
def rank_sent_Pagerank(sim_mat,sentences_list,n=3,alpha=0.85,tol=1.0e-6):
import networkx as nx
nx_graph = nx.from_numpy_array(sim_mat)
scores = nx.pagerank(nx_graph,alpha=alpha,tol=tol)
ranked_sent = sorted(((scores[i],s) for i,s in enumerate(sentences_list)), reverse=True)
ranked_sentences_list = [ tuple_[1] for i, tuple_ in enumerate(ranked_sent)]
return ranked_sentences_list
def get_first_doc(line, tag="|||||"):
first_doc = line.split(tag)[0]
return first_doc
def truncate_doc(doc):
sent_list = sent_detector.tokenize(doc.strip())
seg=6
if len(sent_list)>seg:
return sent_list[:seg]
else:
return sent_list
def read_lead_sentences(doc,tag="story_separator_special_tag"):
doc_list = doc.split(tag)
# print(doc_list)
sent_list=[]
# only keep the first three sources
if len(doc_list) == 2:
for count, doc in enumerate(doc_list):
doc_sent_list = sent_detector.tokenize(doc.strip())
try:
sent_list += doc_sent_list[:4]
except:
sent_list += doc_sent_list
elif len(doc_list) > 2:
doc_list = doc_list[:3]
for count, doc in enumerate(doc_list):
doc_sent_list = sent_detector.tokenize(doc.strip())
try:
sent_list += doc_sent_list[:3]
except:
sent_list += doc_sent_list
print(f"number of sources in this instance: {len(doc_list)}")
print(f"number of lead sentences in this instance: {len(sent_list)}")
print("*"*80)
return sent_list
# read source file into a list of list:
def read_file(path, file_name, read_lead_only=False, read_first_doc=False):
f = open(os.path.join(path, file_name),"r")
lines = f.readlines()
src_list = []
tag="story_separator_special_tag"
for line in lines:
if read_first_doc:
line = get_first_doc(line)
sent_list = truncate_doc(line)
elif read_lead_only:
sent_list = read_lead_sentences(line,tag=tag)
else:
# remove tag; uncomment below for baseline
line = line.replace(tag, "")
# tokenzie line to sentences
sent_list = sent_detector.tokenize(line.strip())
src_list.append(sent_list)
return src_list
def tag_pos(str_text):
doc=spacynlp(str_text)
textlist=[]
# compare the words between two strings
for item in doc:
source_token = item.text
source_pos = item.tag_
textlist.append(source_token+'/'+source_pos)
return ' '.join(textlist)
def convert_sents_to_tagged_sents(sent_list):
tagged_list = []
if(len(sent_list)>0):
for s in sent_list:
s = s.replace("/", "")
# print("original sent -------- \n",s)
temp_tagged = tag_pos(s)
tagged_list.append(temp_tagged)
else:
tagged_list.append(tag_pos("."))
return tagged_list
def get_compressed_sen(sentences, nb_words):
compresser = takahe.word_graph(sentences, nb_words = nb_words, lang = 'en', punct_tag = "." )
candidates = compresser.get_compression(3)
# print("--------------------Top 3 candicate---------------", candidates)
reranker = takahe.keyphrase_reranker(sentences,
candidates,
lang = 'en')
# print("reranker: ", reranker)
# print("finish initialising reranker------------")
reranked_candidates = reranker.rerank_nbest_compressions()
# print(reranked_candidates)
if len(reranked_candidates)>0:
score, path = reranked_candidates[0]
result = ' '.join([u[0] for u in path])
else:
result=' '
# print("----------------selected candicate as final output-------------- ", result)
return result