urls_high_matches = {
    "blog.libero.it": "some articles are erotic Poetry",
    "bust.com": "sex education, non porn type of content",
    "chicagoreader.com": "news articles",
    "discord.com": "tech blogs",
    "filedn.com": "cloud storage, web hosting type",
    "hotnessrater.com": "can't access",
    "ibb.co": "image file hosting",
    "imgur.com": "youtube shots type website",
    "izismile.com": "adult wedsite",
    "jungefreiheit.de": "german news articles",
    "marktplaza.nl": "resale marketplace",
    "pbase.com": "file hosting",
    "racejack.s40.xrea.com": "gambling",
    "rapidgator.net": "file sharing",
    "rutube.ru": "porn",
    "servimg.com": "image hsting",
    "telegra.ph": "some text hosting website",
    "thechive.com": "porn",
    "=3.com": "can't access",
    "thoughtcatalog.com": "Magazine",
    "turbobit.net": "file hosting",
    "videa.hu": "videos like youtube",
    "weheartit.com": "controversy new articles",
    "xnxx.com.se": "porn",
}

urls_false_positives = {
    "bust.com": "sex education, non porn type of content",
    "chicagoreader.com": "news articles",
    "discord.com": "tech blogs",
    "jungefreiheit.de": "german news articles",
    "marktplaza.nl": "resale marketplace",
    "telegra.ph": "some text hosting website",
}