diff --git a/main.py b/main.py index 8dff45d..ff8dbcf 100644 --- a/main.py +++ b/main.py @@ -17,8 +17,8 @@ logger = logging.getLogger(__name__) SOURCE_FILE = './resources/sources.txt' STOPWORDS_FILE = './resources/stopwords.txt' EXCLUDED_PHRASES_FILE = './resources/excluded_phrases.txt' -SIMILARITY_THRESHOLD = 0.75 -MIN_SOURCES = 2 +SIMILARITY_THRESHOLD = 0.68 +MIN_SOURCES = 3 # Used so stories without a parsable date sort behind dated ones. _MIN_DATE = date.min diff --git a/resources/sources.txt b/resources/sources.txt index d48d2dd..ad9d25b 100644 --- a/resources/sources.txt +++ b/resources/sources.txt @@ -25,11 +25,16 @@ https://www.france24.com/en/rss https://www.euronews.com/rss?format=mrss&level=theme&name=news https://www.spiegel.de/international/index.rss https://feeds.elpais.com/mrss-s/pages/ep/site/english.elpais.com/portada +https://www.rt.com/rss/news/ +https://www.euronews.com/news/americas +https://tvpworld.com/69471092/news # Middle East & Asia https://www.aljazeera.com/xml/rss/all.xml https://www.japantimes.co.jp/feed/ -https://www.scmp.com/rss/91/feed +https://www.scmp.com/rss/322262/feed/ +https://www.philstar.com/rss/world +https://www.thejakartapost.com/ # Policy & national https://feeds.washingtonpost.com/rss/world @@ -37,4 +42,9 @@ https://thehill.com/feed/ https://www.vox.com/rss/index.xml # Business -https://www.cnbc.com/id/100003114/device/rss/rss.html \ No newline at end of file +https://www.cnbc.com/id/100003114/device/rss/rss.html + +# Miscellaneous +https://theintercept.com/feed/ +https://www.occrp.org/en +https://www.bellingcat.com/ \ No newline at end of file