skander101 commited on
Commit
3c9e07f
·
1 Parent(s): c10d739

Replace World Health feeds: STAT, ScienceDaily, Science News, Medscape, NIH, Harvard Gazette

Browse files
Files changed (3) hide show
  1. config.py +5 -2
  2. src/analyzer.py +10 -0
  3. src/rss_feed_scraper.py +5 -2
config.py CHANGED
@@ -20,10 +20,13 @@ class Config:
20
  "https://www.aljazeera.com/xml/rss/all.xml",
21
  "https://www.theguardian.com/world/rss",
22
  # ---- World Health ----
23
- "https://feeds.bbci.co.uk/news/health/rss.xml",
24
- "https://rss.nytimes.com/services/xml/rss/nyt/Science.xml",
25
  "https://www.statnews.com/feed/",
26
  "https://www.sciencedaily.com/rss/all.xml",
 
 
 
 
 
27
  # ---- Tech ----
28
  "https://feeds.bbci.co.uk/news/technology/rss.xml",
29
  "https://rss.nytimes.com/services/xml/rss/nyt/Technology.xml",
 
20
  "https://www.aljazeera.com/xml/rss/all.xml",
21
  "https://www.theguardian.com/world/rss",
22
  # ---- World Health ----
 
 
23
  "https://www.statnews.com/feed/",
24
  "https://www.sciencedaily.com/rss/all.xml",
25
+ "https://www.sciencenews.org/feed",
26
+ "https://www.medscape.com/cx/rssfeeds/2700.xml",
27
+ "https://www.nih.gov/news-events/news-releases/rss.xml",
28
+ "https://news.harvard.edu/gazette/feed/",
29
+ "https://news.harvard.edu/gazette/section/health-medicine/feed/",
30
  # ---- Tech ----
31
  "https://feeds.bbci.co.uk/news/technology/rss.xml",
32
  "https://rss.nytimes.com/services/xml/rss/nyt/Technology.xml",
src/analyzer.py CHANGED
@@ -13,6 +13,8 @@ RELIABLE_DOMAINS = {
13
  "economist.com": 0.15, "nature.com": 0.20, "science.org": 0.18,
14
  "sciencedaily.com": 0.14, "theguardian.com": 0.08, "nytimes.com": 0.10,
15
  "washingtonpost.com": 0.10, "ft.com": 0.14, "bloomberg.com": 0.12,
 
 
16
  }
17
 
18
  UNRELIABLE_DOMAINS = {
@@ -73,6 +75,10 @@ SPONSOR_INFO: dict[str, dict] = {
73
  "thedailymash.co.uk": {"display": "Daily Mash", "parent": "Independent", "category": "independent", "bias": "satire", "factuality": "satire", "owners": ["Neil Rafferty (founder)"], "wikipedia": "https://en.wikipedia.org/wiki/The_Daily_Mash"},
74
  "babylonbee.com": {"display": "Babylon Bee", "parent": "Independent", "category": "independent", "bias": "satire", "factuality": "satire", "owners": ["Seth Dillon (CEO)"], "wikipedia": "https://en.wikipedia.org/wiki/Babylon_Bee"},
75
  "sciencedaily.com": {"display": "ScienceDaily", "parent": "ScienceDaily LLC", "category": "independent", "bias": "center", "factuality": "high", "owners": ["Dan Hogan (founder)"], "wikipedia": "https://en.wikipedia.org/wiki/ScienceDaily"},
 
 
 
 
76
  "middleeasteye.net": {"display": "Middle East Eye", "parent": "Independent", "category": "independent", "bias": "left-center", "factuality": "mixed", "owners": ["Independent (London-based, reader-supported)"], "wikipedia": "https://en.wikipedia.org/wiki/Middle_East_Eye"},
77
  "newarab.com": {"display": "The New Arab", "parent": "Fikra Publishing", "category": "independent", "bias": "left-center", "factuality": "mixed", "owners": ["Fikra Publishing Ltd."], "wikipedia": "https://en.wikipedia.org/wiki/The_New_Arab"},
78
  "tunisiaonlinenews.com": {"display": "Tunisia Online News", "parent": "Independent", "category": "independent", "bias": "center", "factuality": "mixed", "owners": ["Independent editorial team"], "wikipedia": ""},
@@ -470,6 +476,10 @@ class NewsAnalyzer:
470
  "deadline.com": "movies",
471
  "screenrant.com": "movies",
472
  "tunisiaonlinenews.com": "tunisia",
 
 
 
 
473
  }
474
 
475
  def _keyword_topic_match(self, article: Article, summary: str) -> list[str]:
 
13
  "economist.com": 0.15, "nature.com": 0.20, "science.org": 0.18,
14
  "sciencedaily.com": 0.14, "theguardian.com": 0.08, "nytimes.com": 0.10,
15
  "washingtonpost.com": 0.10, "ft.com": 0.14, "bloomberg.com": 0.12,
16
+ "sciencenews.org": 0.14, "medscape.com": 0.12, "nih.gov": 0.16,
17
+ "news.harvard.edu": 0.14,
18
  }
19
 
20
  UNRELIABLE_DOMAINS = {
 
75
  "thedailymash.co.uk": {"display": "Daily Mash", "parent": "Independent", "category": "independent", "bias": "satire", "factuality": "satire", "owners": ["Neil Rafferty (founder)"], "wikipedia": "https://en.wikipedia.org/wiki/The_Daily_Mash"},
76
  "babylonbee.com": {"display": "Babylon Bee", "parent": "Independent", "category": "independent", "bias": "satire", "factuality": "satire", "owners": ["Seth Dillon (CEO)"], "wikipedia": "https://en.wikipedia.org/wiki/Babylon_Bee"},
77
  "sciencedaily.com": {"display": "ScienceDaily", "parent": "ScienceDaily LLC", "category": "independent", "bias": "center", "factuality": "high", "owners": ["Dan Hogan (founder)"], "wikipedia": "https://en.wikipedia.org/wiki/ScienceDaily"},
78
+ "sciencenews.org": {"display": "Science News", "parent": "Society for Science", "category": "independent", "bias": "center", "factuality": "high", "owners": ["Society for Science (non-profit)"], "wikipedia": "https://en.wikipedia.org/wiki/Science_News"},
79
+ "medscape.com": {"display": "Medscape", "parent": "WebMD Health Corp", "category": "corporate", "bias": "center", "factuality": "high", "owners": ["Internet Brands (KKR & Co.)"], "wikipedia": "https://en.wikipedia.org/wiki/Medscape"},
80
+ "nih.gov": {"display": "National Institutes of Health", "parent": "U.S. Department of Health and Human Services", "category": "government", "bias": "center", "factuality": "high", "owners": ["U.S. Federal Government"], "wikipedia": "https://en.wikipedia.org/wiki/National_Institutes_of_Health"},
81
+ "news.harvard.edu": {"display": "Harvard Gazette", "parent": "Harvard University", "category": "independent", "bias": "center", "factuality": "high", "owners": ["Harvard University (non-profit educational institution)"], "wikipedia": "https://en.wikipedia.org/wiki/Harvard_Gazette"},
82
  "middleeasteye.net": {"display": "Middle East Eye", "parent": "Independent", "category": "independent", "bias": "left-center", "factuality": "mixed", "owners": ["Independent (London-based, reader-supported)"], "wikipedia": "https://en.wikipedia.org/wiki/Middle_East_Eye"},
83
  "newarab.com": {"display": "The New Arab", "parent": "Fikra Publishing", "category": "independent", "bias": "left-center", "factuality": "mixed", "owners": ["Fikra Publishing Ltd."], "wikipedia": "https://en.wikipedia.org/wiki/The_New_Arab"},
84
  "tunisiaonlinenews.com": {"display": "Tunisia Online News", "parent": "Independent", "category": "independent", "bias": "center", "factuality": "mixed", "owners": ["Independent editorial team"], "wikipedia": ""},
 
476
  "deadline.com": "movies",
477
  "screenrant.com": "movies",
478
  "tunisiaonlinenews.com": "tunisia",
479
+ "sciencenews.org": "health",
480
+ "medscape.com": "health",
481
+ "nih.gov": "health",
482
+ "news.harvard.edu": "health",
483
  }
484
 
485
  def _keyword_topic_match(self, article: Article, summary: str) -> list[str]:
src/rss_feed_scraper.py CHANGED
@@ -21,10 +21,13 @@ DEFAULT_FEEDS = [
21
  "https://www.aljazeera.com/xml/rss/all.xml",
22
  "https://www.theguardian.com/world/rss",
23
  # World Health
24
- "https://feeds.bbci.co.uk/news/health/rss.xml",
25
- "https://rss.nytimes.com/services/xml/rss/nyt/Science.xml",
26
  "https://www.statnews.com/feed/",
27
  "https://www.sciencedaily.com/rss/all.xml",
 
 
 
 
 
28
  # Tech
29
  "https://feeds.bbci.co.uk/news/technology/rss.xml",
30
  "https://rss.nytimes.com/services/xml/rss/nyt/Technology.xml",
 
21
  "https://www.aljazeera.com/xml/rss/all.xml",
22
  "https://www.theguardian.com/world/rss",
23
  # World Health
 
 
24
  "https://www.statnews.com/feed/",
25
  "https://www.sciencedaily.com/rss/all.xml",
26
+ "https://www.sciencenews.org/feed",
27
+ "https://www.medscape.com/cx/rssfeeds/2700.xml",
28
+ "https://www.nih.gov/news-events/news-releases/rss.xml",
29
+ "https://news.harvard.edu/gazette/feed/",
30
+ "https://news.harvard.edu/gazette/section/health-medicine/feed/",
31
  # Tech
32
  "https://feeds.bbci.co.uk/news/technology/rss.xml",
33
  "https://rss.nytimes.com/services/xml/rss/nyt/Technology.xml",