-
Notifications
You must be signed in to change notification settings - Fork 1
/
model.py
266 lines (210 loc) · 10.1 KB
/
model.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
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
import json
import os
from urllib.parse import urlparse
from sentence_transformers import SentenceTransformer, util
import google.generativeai as genai
from dotenv import load_dotenv
from transformers import pipeline
import numpy as np
import spacy
import numpy as np
from sklearn.metrics import precision_score, recall_score, f1_score
import time
def is_webpage_url(url: str) -> bool:
parsed_url = urlparse(url)
if parsed_url.fragment:
return False
file_extensions = (
'.pdf', '.doc', '.docx', '.xls', '.xlsx', '.ppt', '.pptx',
'.jpg', '.jpeg', '.png', '.gif', '.svg', '.bmp', '.tiff',
'.zip', '.rar', '.7z', '.tar', '.gz', '.mp3', '.mp4', '.avi',
'.mov', '.mkv', '.wmv', '.exe', '.dmg', '.iso', '.apk'
)
if any(parsed_url.path.lower().endswith(ext) for ext in file_extensions):
return False
return True
# Load environment variables
load_dotenv()
api_key = os.getenv("API_KEY")
genai.configure(api_key=api_key)
# Load content from JSON file
def load_content(filename='webpage_content.json'):
with open(filename, 'r', encoding='utf-8') as f:
try:
content_list = json.load(f)
print(f"Successfully loaded {len(content_list)} items from {filename}")
return content_list
except json.JSONDecodeError as e:
print(f"Error decoding JSON: {e}")
print("Attempting to read file line by line...")
f.seek(0) # Reset file pointer to the beginning
content_list = []
for line_number, line in enumerate(f, 1):
try:
content = json.loads(line.strip())
content_list.append(content)
except json.JSONDecodeError as e:
print(f"Error decoding JSON on line {line_number}: {e}")
print(f"Problematic line: {line[:100]}...")
print(f"Successfully loaded {len(content_list)} items from {filename}")
return content_list
# Generate questions using Generative AI model
def generate_questions(content):
model = genai.GenerativeModel('gemini-pro')
prompt = f"""
Generate 10 relevant questions from the following content:
{content}
Please ensure the questions are clear, concise, and relevant to the content provided.
"""
response = model.generate_content(prompt)
questions = response.text.strip().split('\n')
return [question.strip() for question in questions if question.strip()]
# Initialize SentenceTransformer model
model = SentenceTransformer('all-MiniLM-L6-v2')
# Find relevant links based on content similarity
def find_relevant_links(content, links, scraped_content_dict):
content_embedding = model.encode(content, convert_to_tensor=True)
relevant_links = []
for link in links:
linked_content = scraped_content_dict.get(link)
if linked_content:
link_embedding = model.encode(linked_content, convert_to_tensor=True)
similarity = util.pytorch_cos_sim(content_embedding, link_embedding).item()
relevant_links.append((link, similarity))
relevant_links.sort(key=lambda x: x[1], reverse=True)
return [link for link, sim in relevant_links[:5]]
# Extract topics from content
def extract_topics(content):
keywords = [word for word in content.split() if word.isalpha()][:5]
return keywords
# Save questions and related data to JSON file
def save_questions_to_json(data, filename='questions_with_content.json'):
try:
mode = 'a' if os.path.exists(filename) else 'w'
with open(filename, mode, encoding='utf-8') as f:
json.dump(data, f, indent=4, ensure_ascii=False)
f.write('\n')
print(f"Successfully saved data for {data['url']} to {filename}")
except Exception as e:
print(f"Error saving data for {data['url']}: {str(e)}")
# Verify the generated questions and links
def verify_data(data):
num_questions = len(data['questions'])
questions_length_valid = all(len(q) <= 80 for q in data['questions'])
num_links = len(data['relevant_links'])
num_topics = len(data['topics'])
if num_questions == 10 and questions_length_valid and num_links == 5 and num_topics == 5:
return True
else:
print(f"Verification failed for {data['url']}:")
print(f"Number of questions: {num_questions} (Expected: 10)")
print(f"All questions under 80 characters: {questions_length_valid}")
print(f"Number of relevant links: {num_links} (Expected: 5)")
print(f"Number of topics: {num_topics} (Expected: 5)")
return False
# Evaluate the relevance of generated questions
def evaluate_question_relevance(generated_questions, content):
content_embedding = model.encode(content, convert_to_tensor=True)
relevance_scores = []
for question in generated_questions:
question_embedding = model.encode(question, convert_to_tensor=True)
similarity = util.pytorch_cos_sim(content_embedding, question_embedding).item()
relevance_scores.append(similarity)
avg_relevance = sum(relevance_scores) / len(relevance_scores) if relevance_scores else 0
return avg_relevance
def load_spacy_model():
try:
return spacy.load("en_core_web_sm")
except OSError:
print("Downloading spaCy model...")
spacy.cli.download("en_core_web_sm")
return spacy.load("en_core_web_sm")
nlp = load_spacy_model()
def evaluate_link_relevance(content, predicted_links, scraped_content_dict):
content_doc = nlp(content[:1000000]) # Limit to first 1M characters to avoid memory issues
relevance_scores = []
for link in predicted_links:
link_content = scraped_content_dict.get(link, "")
link_doc = nlp(link_content[:1000000]) # Limit to first 1M characters
similarity = content_doc.similarity(link_doc)
relevance_scores.append(similarity)
# Normalize scores
total_score = sum(relevance_scores)
normalized_scores = [score / total_score if total_score > 0 else 0 for score in relevance_scores]
# Calculate a weighted relevance score
weighted_relevance = sum(score * (1 / (i + 1)) for i, score in enumerate(normalized_scores))
return weighted_relevance
def load_spacy_model():
try:
return spacy.load("en_core_web_sm")
except OSError:
print("Downloading spaCy model...")
spacy.cli.download("en_core_web_sm")
return spacy.load("en_core_web_sm")
nlp = load_spacy_model()
def evaluate_link_relevance(content, predicted_links, scraped_content_dict):
content_doc = nlp(content[:100000]) # Limit to first 1M characters to avoid memory issues
relevance_scores = []
for link in predicted_links:
link_content = scraped_content_dict.get(link, "")
link_doc = nlp(link_content[:100000]) # Limit to first 1M characters
similarity = content_doc.similarity(link_doc)
relevance_scores.append(similarity)
# Normalize scores
total_score = sum(relevance_scores)
normalized_scores = [score / total_score if total_score > 0 else 0 for score in relevance_scores]
# Calculate a weighted relevance score
weighted_relevance = sum(score * (1 / (i + 1)) for i, score in enumerate(normalized_scores))
return weighted_relevance
# Update the process_content_for_questions function
def process_content_for_questions(content_list, scraped_content_dict, num_urls=5):
links = list(scraped_content_dict.keys())
for entry in content_list[:num_urls]:
try:
content = entry['content']
url = entry['url']
print(f"Processing {url}")
# Truncate content to the first 512 tokens
truncated_content = ' '.join(content.split()[:512])
questions = generate_questions(truncated_content)
print(f"Generated {len(questions)} questions for {url}")
relevant_links = find_relevant_links(truncated_content, links, scraped_content_dict)
print(f"Found {len(relevant_links)} relevant links for {url}")
topics = extract_topics(truncated_content)
print(f"Extracted {len(topics)} topics for {url}")
question_relevance = evaluate_question_relevance(questions, truncated_content)
link_relevance = evaluate_link_relevance(truncated_content, relevant_links, scraped_content_dict)
print("Sleeping for 10 sec to avoid rate limit.")
time.sleep(10)
data = {
"url": url,
"content": truncated_content[:500],
"questions": questions,
"relevant_links": relevant_links,
"topics": topics,
"question_relevance_score": question_relevance,
"link_relevance_score": link_relevance
}
if verify_data(data):
save_questions_to_json(data)
else:
print(f"Skipping saving data for {url} due to verification failure.")
except Exception as e:
print(f"Error processing {url}: {str(e)}")
# Main function to execute the process
if __name__ == "__main__":
try:
content_list = load_content()
print(f"Loaded {len(content_list)} content entries")
if not content_list:
print("Error: No content loaded. Check your input file.")
else:
scraped_content_dict = {entry['url']: entry['content'] for entry in content_list}
print(f"Created scraped_content_dict with {len(scraped_content_dict)} entries")
process_content_for_questions(content_list, scraped_content_dict, num_urls=5)
if os.path.exists('questions_with_content.json'):
print(f"Output file 'questions_with_content.json' has been created and its size is {os.path.getsize('questions_with_content.json')} bytes")
else:
print("Output file 'questions_with_content.json' was not created")
except Exception as e:
print(f"An error occurred: {str(e)}")