Rich sponsor ownership: shareholders, types, ownership chains

refresh_task.py

@@ -55,7 +55,8 @@ def serialize_clusters(clusters):
                     "summary": a.analysis.summary if a.analysis else "",
                     "topics": a.analysis.topics if a.analysis else [],
                     "trust": f"{a.analysis.trustworthiness_score:.0%}" if a.analysis else "",
-                    "sponsor": a.analysis.sponsor if a.analysis else "",
+                    "sponsor": a.analysis.sponsor.get("display", "") if a.analysis else "",
+                    "sponsor_info": a.analysis.sponsor if a.analysis else {},
                     "score": a.post.score,
                     "comments": a.post.num_comments,
                     "url": a.post.url,
@@ -85,7 +86,8 @@ def serialize_clusters(clusters):
                     "summary": a.analysis.summary if a.analysis else "",
                     "topics": a.analysis.topics if a.analysis else [],
                     "trust": round(a.analysis.trustworthiness_score, 2) if a.analysis else 0,
-                    "sponsor": a.analysis.sponsor if a.analysis else "",
+                    "sponsor": a.analysis.sponsor.get("display", "") if a.analysis else "",
+                    "sponsor_info": a.analysis.sponsor if a.analysis else {},
                     "score": a.post.score,
                     "comments": a.post.num_comments,
                     "image": a.article.image_url if a.article else a.post.image_url,

src/analyzer.py

@@ -20,54 +20,53 @@ UNRELIABLE_DOMAINS = {
     "theonion.com": -0.25, "naturalnews.com": -0.30, "zerohedge.com": -0.15,
 }
 
-SPONSORS: dict[str, str] = {
-    "bbc.com": "BBC (publicly funded)",
-    "bbc.co.uk": "BBC (publicly funded)",
-    "nytimes.com": "The New York Times Company",
-    "npr.org": "NPR (publicly funded)",
-    "aljazeera.com": "Al Jazeera Media Network (Qatar-funded)",
-    "theguardian.com": "Guardian Media Group",
-    "statnews.com": "Boston Globe Media",
-    "sciencedaily.com": "ScienceDaily LLC",
-    "techcrunch.com": "Yahoo (Apollo Global Management)",
-    "wired.com": "Condé Nast (Advance Publications)",
-    "theverge.com": "Vox Media",
-    "arstechnica.com": "Condé Nast (Advance Publications)",
-    "krebsonsecurity.com": "Independent (Brian Krebs)",
-    "bleepingcomputer.com": "Independent",
-    "threatpost.com": "Independent",
-    "therecord.media": "Recorded Future",
-    "theonion.com": "Global Tetrahedron",
-    "thedailymash.co.uk": "Independent",
-    "babylonbee.com": "Independent",
-    "ign.com": "Ziff Davis",
-    "eurogamer.net": "ReedPop (Gamer Network)",
-    "pcgamer.com": "Future plc",
-    "kotaku.com": "G/O Media",
-    "gamespot.com": "Fandom",
-    "polygon.com": "Vox Media",
-    "variety.com": "Penske Media Corporation",
-    "hollywoodreporter.com": "Penske Media Corporation",
-    "deadline.com": "Penske Media Corporation",
-    "screenrant.com": "Valnet Inc.",
-    "arabnews.com": "Saudi Research & Publishing",
-    "middleeasteye.net": "Independent (London-based)",
-    "newarab.com": "Fikra Publishing",
-    "france24.com": "France Médias Monde (government-funded)",
-    "tunisiaonlinenews.com": "Independent",
-    "northafricapost.com": "Independent",
-    "africanews.com": "Independent",
-    "reuters.com": "Thomson Reuters",
-    "apnews.com": "Associated Press (non-profit)",
-    "wsj.com": "Dow Jones (News Corp)",
-    "economist.com": "The Economist Group",
-    "nature.com": "Springer Nature",
-    "science.org": "American Association for the Advancement of Science",
-    "washingtonpost.com": "The Washington Post (Jeff Bezos)",
-    "ft.com": "Financial Times (Nikkei)",
-    "bloomberg.com": "Bloomberg L.P. (Michael Bloomberg)",
-    "thehackernews.com": "Independent (THN)",
-    "ign.com": "Ziff Davis",
+SPONSOR_INFO: dict[str, dict] = {
+    "bbc.com": {"display": "BBC (publicly funded)", "parent": "BBC", "type": "state", "owners": ["UK Government (licence fee payers)"]},
+    "bbc.co.uk": {"display": "BBC (publicly funded)", "parent": "BBC", "type": "state", "owners": ["UK Government (licence fee payers)"]},
+    "nytimes.com": {"display": "The New York Times Company", "parent": "The New York Times Company", "type": "public", "owners": ["Ochs-Sulzberger family (controlling)", "Public shareholders (NYSE: NYT)"]},
+    "npr.org": {"display": "NPR (publicly funded)", "parent": "National Public Radio", "type": "nonprofit", "owners": ["Corporation for Public Broadcasting", "Member stations", "Listeners/donors"]},
+    "aljazeera.com": {"display": "Al Jazeera Media Network", "parent": "Al Jazeera Media Network", "type": "state", "owners": ["Government of Qatar"]},
+    "theguardian.com": {"display": "Guardian Media Group", "parent": "Guardian Media Group", "type": "nonprofit", "owners": ["Scott Trust Limited (no shareholders)"]},
+    "statnews.com": {"display": "Boston Globe Media", "parent": "Boston Globe Media Partners", "type": "private", "owners": ["John W. Henry (also owns Boston Red Sox)"]},
+    "sciencedaily.com": {"display": "ScienceDaily LLC", "parent": "ScienceDaily LLC", "type": "private", "owners": ["Dan Hogan (founder)"]},
+    "techcrunch.com": {"display": "TechCrunch (Yahoo)", "parent": "Yahoo Inc.", "type": "private", "owners": ["Apollo Global Management"]},
+    "wired.com": {"display": "Wired (Condé Nast)", "parent": "Condé Nast", "type": "private", "owners": ["Advance Publications", "Newhouse family"]},
+    "theverge.com": {"display": "The Verge (Vox Media)", "parent": "Vox Media", "type": "private", "owners": ["NBCUniversal (Comcast)", "Accel Partners", "General Atlantic"]},
+    "arstechnica.com": {"display": "Ars Technica (Condé Nast)", "parent": "Condé Nast", "type": "private", "owners": ["Advance Publications", "Newhouse family"]},
+    "krebsonsecurity.com": {"display": "Krebs on Security", "parent": "Independent", "type": "independent", "owners": ["Brian Krebs (independent journalist)"]},
+    "bleepingcomputer.com": {"display": "BleepingComputer", "parent": "Independent", "type": "independent", "owners": ["Lawrence Abrams (founder)"]},
+    "threatpost.com": {"display": "Threatpost", "parent": "Independent", "type": "independent", "owners": ["Independent editorial team"]},
+    "therecord.media": {"display": "The Record (Recorded Future)", "parent": "Recorded Future", "type": "private", "owners": ["Insikt Group", "Mastercard (parent company)"]},
+    "theonion.com": {"display": "The Onion", "parent": "Global Tetrahedron", "type": "private", "owners": ["Global Tetrahedron LLC"]},
+    "thedailymash.co.uk": {"display": "Daily Mash", "parent": "Independent", "type": "independent", "owners": ["Neil Rafferty (founder)"]},
+    "babylonbee.com": {"display": "Babylon Bee", "parent": "Independent", "type": "independent", "owners": ["Seth Dillon (CEO)"]},
+    "ign.com": {"display": "IGN (Ziff Davis)", "parent": "Ziff Davis", "type": "public", "owners": ["Public shareholders (NASDAQ: ZD)"]},
+    "eurogamer.net": {"display": "Eurogamer (ReedPop)", "parent": "ReedPop (Gamer Network)", "type": "public", "owners": ["RELX Group (parent of Reed Elsevier)", "Public shareholders"]},
+    "pcgamer.com": {"display": "PC Gamer (Future plc)", "parent": "Future plc", "type": "public", "owners": ["Public shareholders (LSE: FUTR)"]},
+    "kotaku.com": {"display": "Kotaku (G/O Media)", "parent": "G/O Media", "type": "private", "owners": ["Great Hill Partners (private equity)"]},
+    "gamespot.com": {"display": "GameSpot (Fandom)", "parent": "Fandom Inc.", "type": "private", "owners": ["TPG Capital", "Integrated Media Co."]},
+    "polygon.com": {"display": "Polygon (Vox Media)", "parent": "Vox Media", "type": "private", "owners": ["NBCUniversal (Comcast)", "Accel Partners", "General Atlantic"]},
+    "variety.com": {"display": "Variety (Penske Media)", "parent": "Penske Media Corporation", "type": "private", "owners": ["Jay Penske (founder/CEO)"]},
+    "hollywoodreporter.com": {"display": "Hollywood Reporter (Penske Media)", "parent": "Penske Media Corporation", "type": "private", "owners": ["Jay Penske (founder/CEO)"]},
+    "deadline.com": {"display": "Deadline (Penske Media)", "parent": "Penske Media Corporation", "type": "private", "owners": ["Jay Penske (founder/CEO)"]},
+    "screenrant.com": {"display": "Screen Rant (Valnet)", "parent": "Valnet Inc.", "type": "private", "owners": ["Valnet Inc. (digital media group)"]},
+    "arabnews.com": {"display": "Arab News", "parent": "Saudi Research & Publishing", "type": "state", "owners": ["Saudi Government (via SRPC)"]},
+    "middleeasteye.net": {"display": "Middle East Eye", "parent": "Independent", "type": "independent", "owners": ["Independent (London-based, reader-supported)"]},
+    "newarab.com": {"display": "The New Arab", "parent": "Fikra Publishing", "type": "private", "owners": ["Fikra Publishing Ltd."]},
+    "france24.com": {"display": "France 24", "parent": "France Médias Monde", "type": "state", "owners": ["French Government (via public funding)"]},
+    "tunisiaonlinenews.com": {"display": "Tunisia Online News", "parent": "Independent", "type": "independent", "owners": ["Independent editorial team"]},
+    "northafricapost.com": {"display": "North Africa Post", "parent": "Independent", "type": "independent", "owners": ["Independent editorial team"]},
+    "africanews.com": {"display": "Africa News", "parent": "Independent", "type": "independent", "owners": ["Independent (NGO-funded)"]},
+    "reuters.com": {"display": "Reuters (Thomson Reuters)", "parent": "Thomson Reuters", "type": "public", "owners": ["The Woodbridge Company (Thomson family, controlling)", "Public shareholders"]},
+    "apnews.com": {"display": "Associated Press", "parent": "Associated Press", "type": "nonprofit", "owners": ["Member newspapers (cooperative, no shareholders)"]},
+    "wsj.com": {"display": "Wall Street Journal (News Corp)", "parent": "Dow Jones (News Corp)", "type": "public", "owners": ["Murdoch family (controlling, via News Corp)", "Public shareholders (NASDAQ: NWSA)"]},
+    "economist.com": {"display": "The Economist", "parent": "The Economist Group", "type": "private", "owners": ["Pearson plc (50%)", "The Rothschild family", "The Economist editorial staff"]},
+    "nature.com": {"display": "Nature (Springer Nature)", "parent": "Springer Nature", "type": "private", "owners": ["Holtzbrinck Publishing Group", "BC Partners"]},
+    "science.org": {"display": "Science (AAAS)", "parent": "American Association for the Advancement of Science", "type": "nonprofit", "owners": ["AAAS membership (non-profit society)"]},
+    "washingtonpost.com": {"display": "The Washington Post", "parent": "The Washington Post", "type": "private", "owners": ["Jeff Bezos (Nash Holdings)"]},
+    "ft.com": {"display": "Financial Times (Nikkei)", "parent": "Financial Times", "type": "public", "owners": ["Nikkei Inc. (Japanese media group)"]},
+    "bloomberg.com": {"display": "Bloomberg", "parent": "Bloomberg L.P.", "type": "private", "owners": ["Michael Bloomberg (88%)", "Merck family (12%)"]},
+    "thehackernews.com": {"display": "The Hacker News", "parent": "Independent", "type": "independent", "owners": ["Independent editorial team"]},
 }
 
 CLICKBAIT_PATTERNS = [
@@ -250,9 +249,12 @@ class NewsAnalyzer:
             logger.warning("Model loading failed: %s — using rule-based", exc)
 
     @staticmethod
-    def _detect_sponsor(article: Article) -> str:
+    def _detect_sponsor(article: Article) -> dict:
         domain = re.sub(r"^www\.", "", (article.source_domain or ""))
-        return SPONSORS.get(domain, "")
+        info = SPONSOR_INFO.get(domain)
+        if info:
+            return dict(info)
+        return {"display": "", "parent": "", "type": "", "owners": []}
 
     def analyze(self, article: Article) -> Analysis:
         summary = self._summarise(article)

src/models.py

@@ -34,7 +34,7 @@ class Analysis:
     is_opinion: bool
     political_leaning: str = "centrist"
     category: str = "General"
-    sponsor: str = ""
+    sponsor: dict = field(default_factory=dict)
 
 
 @dataclass

templates/sponsors.html

@@ -23,7 +23,7 @@ body {
   backdrop-filter: blur(12px);
 }
 .header-inner {
-  max-width: 1400px;
+  max-width: 1200px;
   margin: 0 auto;
   display: flex;
   align-items: center;
@@ -38,6 +38,7 @@ body {
   -webkit-background-clip: text;
   -webkit-text-fill-color: transparent;
 }
+.subtitle { font-size: 13px; color: #64748b; }
 .back-btn {
   padding: 8px 18px;
   border-radius: 8px;
@@ -49,26 +50,46 @@ body {
   transition: background .15s;
 }
 .back-btn:hover { background: #475569; color: #e2e8f0; }
-.container { max-width: 1000px; margin: 0 auto; padding: 32px; }
+.container { max-width: 900px; margin: 0 auto; padding: 32px; }
 .grid { display: grid; gap: 16px; }
 .card {
   background: #1e293b;
   border-radius: 14px;
   padding: 20px 24px;
   border: 1px solid #334155;
+  transition: border-color .15s;
 }
-.card h2 {
-  font-size: 16px;
+.card:hover { border-color: #fbbf24; }
+.card-header { display: flex; align-items: center; justify-content: space-between; margin-bottom: 8px; flex-wrap: wrap; gap: 8px; }
+.card h2 { font-size: 16px; font-weight: 600; color: #fbbf24; }
+.type-badge {
+  font-size: 10px;
   font-weight: 600;
-  color: #fbbf24;
-  margin-bottom: 4px;
+  text-transform: uppercase;
+  letter-spacing: .08em;
+  padding: 3px 8px;
+  border-radius: 5px;
 }
-.card .sub {
+.type-public { background: rgba(59,130,246,0.15); color: #60a5fa; }
+.type-private { background: rgba(251,191,36,0.15); color: #fbbf24; }
+.type-state { background: rgba(239,68,68,0.15); color: #f87171; }
+.type-nonprofit { background: rgba(34,197,94,0.15); color: #4ade80; }
+.type-independent { background: rgba(148,163,184,0.15); color: #94a3b8; }
+.card .parent { font-size: 13px; color: #94a3b8; margin-bottom: 12px; }
+.owners { margin-bottom: 12px; }
+.owners h3 { font-size: 11px; font-weight: 600; color: #64748b; text-transform: uppercase; letter-spacing: .06em; margin-bottom: 6px; }
+.owner-list { display: flex; flex-wrap: wrap; gap: 6px; }
+.owner-tag {
   font-size: 12px;
-  color: #64748b;
-  margin-bottom: 12px;
+  padding: 4px 10px;
+  border-radius: 6px;
+  background: rgba(251,191,36,0.08);
+  color: #fbbf24;
+  border: 1px solid rgba(251,191,36,0.2);
 }
-.sources { display: flex; flex-wrap: wrap; gap: 6px; }
+.sources { margin-top: 12px; padding-top: 12px; border-top: 1px solid #334155; }
+.sources h3 { font-size: 11px; font-weight: 600; color: #64748b; text-transform: uppercase; letter-spacing: .06em; margin-bottom: 6px; }
+.source-list { display: flex; flex-wrap: wrap; gap: 6px; }
 .source-tag {
   font-size: 12px;
   padding: 4px 10px;
@@ -79,14 +100,6 @@ body {
 }
 .empty { text-align: center; padding: 60px 20px; color: #64748b; }
 .empty h2 { font-size: 18px; margin-bottom: 6px; }
-.count-badge {
-  font-size: 11px;
-  color: #818cf8;
-  background: rgba(129, 140, 248, 0.12);
-  padding: 2px 8px;
-  border-radius: 5px;
-  margin-left: 8px;
-}
 @media (max-width: 640px) {
   .header { padding: 14px; }
   .container { padding: 14px; }
@@ -96,7 +109,10 @@ body {
 <body>
 <div class="header">
   <div class="header-inner">
-    <h1>News Digest — Sponsors</h1>
+    <div>
+      <h1>🏢 Who Owns the News?</h1>
+      <div class="subtitle">Track the parent companies and shareholders behind each news source</div>
+    </div>
     <a href="/" class="back-btn">← Back to News</a>
   </div>
 </div>
@@ -105,12 +121,28 @@ body {
   <div class="grid">
     {% for s in sponsors %}
     <div class="card">
-      <h2>🏢 {{ s.name }}</h2>
-      <div class="sub">{{ s.source_count }} source{% if s.source_count != 1 %}s{% endif %}</div>
+      <div class="card-header">
+        <h2>{{ s.display }}</h2>
+        <span class="type-badge type-{{ s.type }}">{{ s.type }}</span>
+      </div>
+      <div class="parent">Parent: {{ s.parent }}</div>
+      {% if s.owners %}
+      <div class="owners">
+        <h3>Shareholders / Funders</h3>
+        <div class="owner-list">
+          {% for owner in s.owners %}
+          <span class="owner-tag">{{ owner }}</span>
+          {% endfor %}
+        </div>
+      </div>
+      {% endif %}
       <div class="sources">
-        {% for domain in s.sources %}
-        <span class="source-tag">{{ domain }}</span>
-        {% endfor %}
+        <h3>{{ s.source_count }} outlet{% if s.source_count != 1 %}s{% endif %} owned</h3>
+        <div class="source-list">
+          {% for domain in s.sources %}
+          <span class="source-tag">{{ domain }}</span>
+          {% endfor %}
+        </div>
       </div>
     </div>
     {% endfor %}
@@ -118,7 +150,7 @@ body {
   {% else %}
   <div class="empty">
     <h2>No sponsor data yet</h2>
-    <p>Run a data refresh to populate sponsor information</p>
+    <p>Run a data refresh to populate ownership information</p>
   </div>
   {% endif %}
 </div>

webapp.py

@@ -55,7 +55,8 @@ def _cluster_to_html_dict(c):
                 "summary": a.analysis.summary if a.analysis else "",
                 "topics": a.analysis.topics if a.analysis else [],
                 "trust": f"{a.analysis.trustworthiness_score:.0%}" if a.analysis else "",
-                "sponsor": a.analysis.sponsor if a.analysis else "",
+                "sponsor": a.analysis.sponsor.get("display", "") if a.analysis else "",
+                "sponsor_info": a.analysis.sponsor if a.analysis else {},
                 "score": a.post.score,
                 "comments": a.post.num_comments,
                 "url": a.post.url,
@@ -90,7 +91,8 @@ def _cluster_to_api_dict(c):
                 "summary": a.analysis.summary if a.analysis else "",
                 "topics": a.analysis.topics if a.analysis else [],
                 "trust": round(a.analysis.trustworthiness_score, 2) if a.analysis else 0,
-                "sponsor": a.analysis.sponsor if a.analysis else "",
+                "sponsor": a.analysis.sponsor.get("display", "") if a.analysis else "",
+                "sponsor_info": a.analysis.sponsor if a.analysis else {},
                 "score": a.post.score,
                 "comments": a.post.num_comments,
                 "image": a.article.image_url if a.article else a.post.image_url,
@@ -227,21 +229,35 @@ def api_categories():
 def _build_sponsors():
     with cache_lock:
         results = list(cached_results)
-    groups: dict[str, set[str]] = {}
+    groups: dict[str, dict] = {}
     for cluster in results:
         for article in cluster.get("articles", []):
-            sponsor = article.get("sponsor", "")
-            domain = article.get("domain", "")
-            if sponsor:
-                if sponsor not in groups:
-                    groups[sponsor] = set()
-                groups[sponsor].add(domain)
+            info = article.get("sponsor_info", {})
+            display = info.get("display", "") or article.get("sponsor", "")
+            if not display:
+                continue
+            if display not in groups:
+                groups[display] = {
+                    "display": display,
+                    "parent": info.get("parent", ""),
+                    "type": info.get("type", ""),
+                    "owners": info.get("owners", []),
+                    "sources": set(),
+                }
+            groups[display]["sources"].add(article.get("domain", ""))
     return sorted(
         [
-            {"name": s, "sources": sorted(doms), "source_count": len(doms)}
-            for s, doms in groups.items()
+            {
+                "display": g["display"],
+                "parent": g["parent"],
+                "type": g["type"],
+                "owners": g["owners"],
+                "sources": sorted(g["sources"]),
+                "source_count": len(g["sources"]),
+            }
+            for g in groups.values()
         ],
-        key=lambda x: x["name"].lower(),
+        key=lambda x: x["display"].lower(),
     )