Align leaderboard values with camera-ready paper

index.html

@@ -839,7 +839,7 @@
           </div>
           <div class="hero-point">
             <strong>Official policy</strong>
-            <span>Official leaderboard entries are validated and judged by the SourceBench team using fixed hidden evaluation settings.</span>
+            <span>Official leaderboard entries are validated and judged by the SourceBench team using a fixed evaluation configuration.</span>
           </div>
         </div>
       </article>
@@ -976,9 +976,9 @@
         </div>
       </div>
       <div class="markdown-note">
-        <p><strong>Local self-check</strong> can be run with the public SourceBench benchmark code and the fixed public query split.</p>
+        <p><strong>Local self-check</strong> can be run with the public SourceBench benchmark code.</p>
         <p><strong>Official leaderboard entries</strong> are not accepted from participant-computed final scores alone. Instead, entries are validated and judged by the SourceBench team.</p>
-        <p>For official evaluation, SourceBench uses hidden holdout queries, the fixed judging setup, and the fixed metric computation pipeline so that leaderboard rows remain comparable across systems.</p>
+        <p>For official evaluation, SourceBench uses a fixed judging setup and metric computation pipeline so that leaderboard rows remain comparable across systems.</p>
       </div>
     </section>
 
@@ -990,7 +990,7 @@
         </div>
       </div>
       <div class="markdown-note">
-        <p><strong>Preferred submission: endpoint access.</strong> Submit the model endpoint, API key, model name, API format, and optional generation settings. The SourceBench team will run hidden queries, source collection, scraping, judging, and metric computation on our side.</p>
+        <p><strong>Preferred submission: endpoint access.</strong> Submit the model endpoint, API key, model name, API format, and optional generation settings. The SourceBench team will run query evaluation, source collection, scraping, judging, and metric computation on our side.</p>
         <p><strong>Fallback submission: answer + cited URL bundle.</strong> If endpoint access cannot be shared, submit per-query answer text together with cited URLs. The SourceBench team will run scraping, post-processing, judging, and metric computation server-side.</p>
         <p><strong>Why these boundaries?</strong> They keep the standardized parts of the benchmark under SourceBench control. Official leaderboard entries are validated and judged by the SourceBench team, rather than accepted from participant-provided final scores.</p>
         <p><strong>Benchmark repository and submission examples:</strong></p>
@@ -1043,9 +1043,9 @@
       freshness: "Freshness",
       objectivity_tone: "Objectivity / Tone",
       layout_ad_density: "Layout / Ad Density",
-      accountability: "Accountability",
-      transparency: "Transparency",
-      authority: "Authority",
+      accountability: "Author Accountability",
+      transparency: "Ownership",
+      authority: "Domain Authority",
     };
 
     const DEEPSEEK_STUDY_MODELS = new Set([

leaderboard_data.json

@@ -1,6 +1,6 @@
 {
   "metadata": {
-    "generated_at": "2026-03-18T04:26:27Z",
+    "generated_at": "2026-11-07T04:23:29Z",
     "scoring_dimensions": [
       "semantic_relevance",
       "factual_accuracy",
@@ -106,7 +106,7 @@
         "rank_file": "data/rank-scores/tavily/geo_scores.json"
       }
     ],
-    "source_of_truth": "Scoring Sheet - Overall-Result-Metric.csv and Scoring Sheet - Overall-Result-Category.csv"
+    "source_of_truth": "Camera-ready paper tables"
   },
   "overall": [
     {
@@ -160,14 +160,14 @@
       "num_complete_scores": 307,
       "unweighted_mean_score": 4.153,
       "weighted_total_content_score": 83.381,
-      "semantic_relevance": 4.264,
-      "factual_accuracy": 4.502,
-      "freshness": 4.192,
-      "objectivity_tone": 4.003,
-      "layout_ad_density": 3.632,
-      "accountability": 4.0,
-      "transparency": 4.3,
-      "authority": 4.329,
+      "semantic_relevance": 4.328,
+      "factual_accuracy": 4.508,
+      "freshness": 4.18,
+      "objectivity_tone": 3.961,
+      "layout_ad_density": 3.633,
+      "accountability": 3.993,
+      "transparency": 4.315,
+      "authority": 4.334,
       "avg_ge_freq": 0.7316051948051939,
       "relative_se_rank": 1.7286246441730713,
       "normalized_reciprocal_se_rank": 0.14088939196167136,
@@ -253,8 +253,8 @@
       "freshness": 4.445,
       "objectivity_tone": 3.898,
       "layout_ad_density": 3.336,
-      "accountability": 4.054,
-      "transparency": 4.163,
+      "accountability": 4.163,
+      "transparency": 4.054,
       "authority": 4.061,
       "avg_ge_freq": null,
       "relative_se_rank": null,
@@ -275,8 +275,8 @@
       "freshness": 4.358,
       "objectivity_tone": 3.619,
       "layout_ad_density": 3.643,
-      "accountability": 3.806,
-      "transparency": 4.22,
+      "accountability": 4.22,
+      "transparency": 3.806,
       "authority": 4.183,
       "avg_ge_freq": 0.8135451187335091,
       "relative_se_rank": 1.6003800744631216,
@@ -291,7 +291,7 @@
       "num_queries": 99,
       "num_complete_scores": 403,
       "unweighted_mean_score": 4.0,
-      "weighted_total_content_score": 79.939,
+      "weighted_total_content_score": 79.903,
       "semantic_relevance": 3.948,
       "factual_accuracy": 4.222,
       "freshness": 4.059,
@@ -319,8 +319,8 @@
       "freshness": 4.346,
       "objectivity_tone": 4.024,
       "layout_ad_density": 3.391,
-      "accountability": 4.125,
-      "transparency": 4.306,
+      "accountability": 4.306,
+      "transparency": 4.125,
       "authority": 4.24,
       "avg_ge_freq": 1.0,
       "relative_se_rank": 1.4974151993912352,
@@ -341,8 +341,8 @@
       "freshness": 4.448,
       "objectivity_tone": 3.801,
       "layout_ad_density": 3.332,
-      "accountability": 4.043,
-      "transparency": 4.033,
+      "accountability": 4.033,
+      "transparency": 4.043,
       "authority": 4.156,
       "avg_ge_freq": 0.9864979746835443,
       "relative_se_rank": 1.2450772341845837,
@@ -358,14 +358,14 @@
       "num_complete_scores": 377,
       "unweighted_mean_score": 4.06,
       "weighted_total_content_score": 81.795,
-      "semantic_relevance": 4.432,
-      "factual_accuracy": 4.426,
-      "freshness": 4.344,
+      "semantic_relevance": 4.435,
+      "factual_accuracy": 4.432,
+      "freshness": 4.343,
       "objectivity_tone": 3.895,
-      "layout_ad_density": 3.34,
-      "accountability": 3.966,
-      "transparency": 4.021,
-      "authority": 4.092,
+      "layout_ad_density": 3.343,
+      "accountability": 3.97,
+      "transparency": 3.97,
+      "authority": 4.102,
       "avg_ge_freq": 0.5340209424083775,
       "relative_se_rank": 1.7669587654960388,
       "normalized_reciprocal_se_rank": 0.1299587450378004,
@@ -379,7 +379,7 @@
       "num_queries": 19,
       "num_complete_scores": 76,
       "unweighted_mean_score": 4.26378842676311,
-      "weighted_total_content_score": 81.1168164313222,
+      "weighted_total_content_score": 85.235,
       "semantic_relevance": 4.243589743589744,
       "factual_accuracy": 4.564102564102564,
       "freshness": 4.423076923076923,
@@ -401,7 +401,7 @@
       "num_queries": 19,
       "num_complete_scores": 58,
       "unweighted_mean_score": 4.282327586206897,
-      "weighted_total_content_score": 80.1018675721562,
+      "weighted_total_content_score": 85.626,
       "semantic_relevance": 4.189655172413793,
       "factual_accuracy": 4.603448275862069,
       "freshness": 4.396551724137931,
@@ -423,7 +423,7 @@
       "num_queries": 19,
       "num_complete_scores": 74,
       "unweighted_mean_score": 4.166666666666667,
-      "weighted_total_content_score": 76.71215074723848,
+      "weighted_total_content_score": 83.286,
       "semantic_relevance": 4.04,
       "factual_accuracy": 4.466666666666667,
       "freshness": 4.351351351351352,
@@ -445,7 +445,7 @@
       "num_queries": 19,
       "num_complete_scores": 69,
       "unweighted_mean_score": 3.963768115942029,
-      "weighted_total_content_score": 72.70175438596492,
+      "weighted_total_content_score": 79.024,
       "semantic_relevance": 3.782608695652174,
       "factual_accuracy": 4.086956521739131,
       "freshness": 4.217391304347826,