Hugging Face's logo

Spaces:
aubynsamuel05
/
nli_checks
Sleeping

App Files Community
nli_checks
File size: 4,847 Bytes 
f078461
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
 
import tldextract
import re
from multiprocessing import Pool, cpu_count
from tqdm import tqdm

# Trusted domains
TRUSTED_DOMAINS = {
    # 🌍 International Mainstream News
    "abcnews.go.com",
    "aljazeera.com",
    "apnews.com",
    "bbc.com",
    "bloomberg.com",
    "cbc.ca",
    "cbsnews.com",
    "cnn.com",
    "dw.com",
    "economist.com",
    "euronews.com",
    "forbes.com",
    "ft.com",
    "indiatimes.com",
    "japantimes.co.jp",
    "latimes.com",
    "npr.org",
    "nytimes.com",
    "reuters.com",
    "smh.com.au",
    "theguardian.com",
    "usatoday.com",
    "washingtonpost.com",
    "wsj.com",
    "france24.com",
    # 📰 Ghana-Specific News
    "3news.com",
    "adomonline.com",
    "citinewsroom.com",
    "ghanaweb.com",
    "ghanaiantimes.com.gh",
    "ghananewsagency.org",
    "graphic.com.gh",
    "modernghana.com",
    "myjoyonline.com",
    "peacefmonline.com",
    "pulse.com.gh",
    "starrfm.com.gh",
    "thebftonline.com",
    "yen.com.gh",
    "nsmq.com.gh",
    # ⚽ Sports News
    "cbssports.com",
    "espn.com",
    "eurosport.com",
    "fifa.com",
    "footballghana.com",
    "foxsports.com",
    "ghanasoccernet.com",
    "goal.com",
    "nba.com",
    "nbcsports.com",
    "onefootball.com",
    "skysports.com",
    "sportinglife.com",
    "supersport.com",
    "tntsports.co.uk",
    "theathletic.com",
    "olympics.com",
    # 🎬 Entertainment & Pop Culture
    "billboard.com",
    "deadline.com",
    "entertainment.com",
    "eonline.com",
    "ew.com",
    "hollywoodreporter.com",
    "indiewire.com",
    "people.com",
    "rollingstone.com",
    "thewrap.com",
    "variety.com",
    # 🧪 Science & Research
    "eurekalert.org",
    "medpagetoday.com",
    "nasa.gov",
    "nature.com",
    "sciencealert.com",
    "sciencenews.org",
    "statnews.com",
    # 🌐 Fact-Checking & Watchdogs
    "africacheck.org",
    "factcheck.org",
    "fullfact.org",
    "politifact.com",
    "snopes.com",
    # 🌍 Global & General Niche News
    "asia.nikkei.com",
    "globalissues.org",
    "ipsnews.net",
    "oecdobserver.org",
    "rferl.org",
    # 📰 African Regional News (non-Ghana)
    "dailynation.africa",
    "enca.com",
    "ewn.co.za",
    "monitor.co.ug",
    "thecitizen.co.tz",
    "businessinsider.com",
    "africanews.com",
    # 🎓 Academic & Policy Think Tanks
    "brookings.edu",
    "carnegieendowment.org",
    "cfr.org",
    "foreignpolicy.com",
    "theconversation.com",
}

# Suspicious domains that often spread misinformation
SUSPICIOUS_DOMAINS = {
    "beforeitsnews.com",
    "naturalnews.com",
    "infowars.com",
    "breitbart.com",
    "dailystormer.com",
    "zerohedge.com",
    "activistpost.com",
    "realfarmacy.com",
    "healthnutnews.com",
}


def extract_domain(url):
    """Extract domain from URL"""
    ext = tldextract.extract(url)
    return f"{ext.domain}.{ext.suffix}"


_PATTERNS = [
    (re.compile(r"\b[A-Z]+\s*\(Reuters\)\s*[-–—]?\s*", re.IGNORECASE), ""),
    (re.compile(r"\(Reuters\)", re.IGNORECASE), ""),
    (re.compile(r"Reuters", re.IGNORECASE), ""),
    (
        re.compile(
            r"\b(?:WASHINGTON|NEW YORK|LONDON|PARIS|BERLIN|TOKYO|MOSCOW|BEIJING|DELHI)\s*[-–—]?\s*",
            re.IGNORECASE,
        ),
        "",
    ),
    (re.compile(r"\b(?:AP|CNN|BBC|Fox News|NBC|CBS|ABC News)\b", re.IGNORECASE), ""),
    (re.compile(r"\bBy\s+[A-Z][a-z]+(?:\s+[A-Z][a-z]+)*\b", re.IGNORECASE), ""),
    (re.compile(r"\S+@\S+\.\S+"), ""),
    (re.compile(r"http[s]?://\S+"), ""),
    (re.compile(r"[^a-zA-Z\s]"), " "),
    (re.compile(r"\s+"), " "),
]


def remove_source_artifacts_fast(text):
    """Optimized version of source artifact removal"""
    if not isinstance(text, str) or len(text) < 10:
        return ""

    for pattern, replacement in _PATTERNS:
        text = pattern.sub(replacement, text)

    return text.strip().lower()


def _process_text_chunk(text_chunk):
    """Internal helper to process a chunk of texts in parallel"""
    return [remove_source_artifacts_fast(text) for text in text_chunk]


def parallel_preprocess(texts, n_jobs=None):
    """Parallel preprocessing of texts using multiprocessing"""
    if n_jobs is None:
        n_jobs = min(cpu_count(), 8)

    chunk_size = max(1, len(texts) // n_jobs)
    chunks = [texts[i : i + chunk_size] for i in range(0, len(texts), chunk_size)]

    print(
        f"Processing {len(texts)} texts in {len(chunks)} chunks using {n_jobs} processes..."
    )

    with Pool(n_jobs) as pool:
        results = list(
            tqdm(
                pool.imap(_process_text_chunk, chunks),
                total=len(chunks),
                desc="Preprocessing chunks",
            )
        )

    processed_texts = []
    for chunk_result in results:
        processed_texts.extend(chunk_result)

    return processed_texts