feat: add Breaking Intelligence v2 calibration

frontend/cloudflare/db/migrations/20260702_breaking_v2_calibration.sql

@@ -0,0 +1,21 @@
+ALTER TABLE events ADD COLUMN breaking_raw_score REAL;
+ALTER TABLE events ADD COLUMN breaking_percentile REAL;
+ALTER TABLE events ADD COLUMN breaking_calibrated_score REAL;
+ALTER TABLE events ADD COLUMN breaking_adversarial_flags TEXT DEFAULT '{}';
+
+UPDATE events
+SET
+    breaking_raw_score = COALESCE(breaking_raw_score, breaking_score),
+    breaking_percentile = COALESCE(breaking_percentile, breaking_score),
+    breaking_calibrated_score = COALESCE(breaking_calibrated_score, breaking_score),
+    breaking_score_version = COALESCE(breaking_score_version, 'breaking-v2.0')
+WHERE breaking_score IS NOT NULL;
+
+DROP INDEX IF EXISTS idx_events_public_breaking;
+CREATE INDEX IF NOT EXISTS idx_events_public_breaking
+ON events(
+    pipeline_stage,
+    breaking_calibrated_score DESC,
+    breaking_score DESC,
+    published_at DESC
+);

frontend/cloudflare/db/schema.sql

@@ -36,11 +36,15 @@ CREATE TABLE IF NOT EXISTS events (
     discussion_count INTEGER,
     classification TEXT DEFAULT '{}',
     extra TEXT DEFAULT '{}',
+    breaking_raw_score REAL,
+    breaking_percentile REAL,
+    breaking_calibrated_score REAL,
     breaking_score REAL,
     breaking_label TEXT,
     breaking_reason TEXT,
     breaking_confidence INTEGER,
     breaking_dimensions TEXT DEFAULT '{}',
+    breaking_adversarial_flags TEXT DEFAULT '{}',
     breaking_score_version TEXT,
     target_timezone TEXT DEFAULT 'UTC',
     published_at_local TEXT,
@@ -56,7 +60,12 @@ CREATE INDEX IF NOT EXISTS idx_events_pipeline_stage ON events(pipeline_stage);
 CREATE INDEX IF NOT EXISTS idx_events_source_id ON events(source_id);
 CREATE INDEX IF NOT EXISTS idx_events_value_label ON events(value_label);
 CREATE INDEX IF NOT EXISTS idx_events_public_featured ON events(pipeline_stage, value_score DESC, published_at DESC);
-CREATE INDEX IF NOT EXISTS idx_events_public_breaking ON events(pipeline_stage, breaking_score DESC, published_at DESC);
+CREATE INDEX IF NOT EXISTS idx_events_public_breaking ON events(
+    pipeline_stage,
+    breaking_calibrated_score DESC,
+    breaking_score DESC,
+    published_at DESC
+);
 
 CREATE TABLE IF NOT EXISTS event_localizations (
     event_id TEXT NOT NULL,

frontend/cloudflare/workers/api/bootstrap.ts

@@ -20,6 +20,7 @@ import {
   publicNewsOrderBy,
   publicNewsLocaleJoin,
   publicNewsSelectColumnsForLocale,
+  pollAfterMsForPublicNews,
   rowToPublicNewsItem,
 } from "../lib/public-news-query";
 import {
@@ -177,11 +178,12 @@ export async function handleBootstrap(
 
     // 构建 news
     const newsRows = (newsResult.results || []) as NewsRow[];
+    const newsItems = newsRows.map((r) => rowToPublicNewsItem(r));
     const newsFeed: PublicNewsFeedResponse = {
-      items: newsRows.map((r) => rowToPublicNewsItem(r)),
+      items: newsItems,
       latestCursor: null,
       nextCursor: null,
-      pollAfterMs: featured ? 30000 : 60000,
+      pollAfterMs: pollAfterMsForPublicNews(featured, newsItems),
       hasNewer: false,
       total: newsCountResult?.total ?? newsRows.length,
     };

frontend/cloudflare/workers/api/news.ts

@@ -18,6 +18,7 @@ import {
   publicNewsOrderBy,
   publicNewsLocaleJoin,
   publicNewsSelectColumnsForLocale,
+  pollAfterMsForPublicNews,
   rowToPublicNewsItem,
 } from "../lib/public-news-query";
 import {
@@ -46,6 +47,7 @@ interface CursorRow {
   published_at: string;
   value_score: number | null;
   breaking_score: number | null;
+  breaking_calibrated_score: number | null;
 }
 
 function withLocaleHeaders(response: Response, locale: string): Response {
@@ -69,22 +71,22 @@ async function buildCursorFilter(
   if (!eventId) return { sql: "", bindings: [] };
 
   const cursor = await db
-    .prepare("SELECT event_id, published_at, value_score, breaking_score FROM events WHERE event_id = ?")
+    .prepare("SELECT event_id, published_at, value_score, breaking_score, breaking_calibrated_score FROM events WHERE event_id = ?")
     .bind(eventId)
     .first<CursorRow>();
   if (!cursor) return { sql: "", bindings: [] };
 
   if (featured) {
-    const score = cursor.breaking_score ?? cursor.value_score ?? -1;
+    const score = cursor.breaking_calibrated_score ?? cursor.breaking_score ?? cursor.value_score ?? -1;
     const scoreOp = mode === "before" ? "<" : ">";
     const timeOp = mode === "before" ? "<" : ">";
     const idOp = mode === "before" ? "<" : ">";
     return {
       sql: `
         AND (
-          COALESCE(breaking_score, value_score, -1) ${scoreOp} ?
-          OR (COALESCE(breaking_score, value_score, -1) = ? AND events.published_at ${timeOp} ?)
-          OR (COALESCE(breaking_score, value_score, -1) = ? AND events.published_at = ? AND events.event_id ${idOp} ?)
+          COALESCE(breaking_calibrated_score, breaking_score, value_score, -1) ${scoreOp} ?
+          OR (COALESCE(breaking_calibrated_score, breaking_score, value_score, -1) = ? AND events.published_at ${timeOp} ?)
+          OR (COALESCE(breaking_calibrated_score, breaking_score, value_score, -1) = ? AND events.published_at = ? AND events.event_id ${idOp} ?)
         )
       `,
       bindings: [score, score, cursor.published_at, score, cursor.published_at, cursor.event_id],
@@ -216,7 +218,7 @@ export async function handleNewsFeed(
       items,
       latestCursor,
       nextCursor,
-      pollAfterMs: featured ? 30000 : 60000,
+      pollAfterMs: pollAfterMsForPublicNews(featured, items),
       hasNewer: Boolean(sinceCursor && items.length > 0),
       total: totalResult?.total ?? rows.length,
     };

frontend/cloudflare/workers/api/webhook.ts

@@ -8,6 +8,7 @@
 
 import type { WebhookResponse, ImportResponse, ImportEventItem } from "../lib/contracts";
 import { internalError } from "../lib/errors";
+import { refreshBreakingScoreStats } from "../lib/breaking-calibration";
 import { refreshPublicReadSnapshots } from "../lib/public-read-snapshots";
 
 type ImportEventWithId = ImportEventItem & {
@@ -66,11 +67,15 @@ function hasProjectionUpdate(item: ImportEventWithId): boolean {
       jsonArrayText(item.image_urls) ||
       jsonArrayText(item.entities) ||
       jsonObjectText(item.classification) ||
+      typeof item.breaking_raw_score === "number" ||
+      typeof item.breaking_percentile === "number" ||
+      typeof item.breaking_calibrated_score === "number" ||
       typeof item.breaking_score === "number" ||
       nonEmptyText(item.breaking_label) ||
       nonEmptyText(item.breaking_reason) ||
       typeof item.breaking_confidence === "number" ||
       jsonObjectText(item.breaking_dimensions) ||
+      jsonObjectText(item.breaking_adversarial_flags) ||
       nonEmptyText(item.target_timezone) ||
       nonEmptyText(item.published_at_local) ||
       (Array.isArray(item.localizations) && item.localizations.length > 0),
@@ -204,11 +209,15 @@ export async function importEventsToD1(
              value_score = COALESCE(?, value_score),
              china_relevance_label = COALESCE(NULLIF(?, ''), china_relevance_label),
              classification = COALESCE(?, classification),
+             breaking_raw_score = COALESCE(?, breaking_raw_score),
+             breaking_percentile = COALESCE(?, breaking_percentile),
+             breaking_calibrated_score = COALESCE(?, breaking_calibrated_score),
              breaking_score = COALESCE(?, breaking_score),
              breaking_label = COALESCE(NULLIF(?, ''), breaking_label),
              breaking_reason = COALESCE(NULLIF(?, ''), breaking_reason),
              breaking_confidence = COALESCE(?, breaking_confidence),
              breaking_dimensions = COALESCE(?, breaking_dimensions),
+             breaking_adversarial_flags = COALESCE(?, breaking_adversarial_flags),
              breaking_score_version = COALESCE(NULLIF(?, ''), breaking_score_version),
              target_timezone = COALESCE(NULLIF(?, ''), target_timezone),
              published_at_local = COALESCE(NULLIF(?, ''), published_at_local),
@@ -231,11 +240,15 @@ export async function importEventsToD1(
           typeof item.value_score === "number" ? item.value_score : null,
           nonEmptyText(item.china_relevance_label),
           jsonObjectText(item.classification),
+          typeof item.breaking_raw_score === "number" ? item.breaking_raw_score : null,
+          typeof item.breaking_percentile === "number" ? item.breaking_percentile : null,
+          typeof item.breaking_calibrated_score === "number" ? item.breaking_calibrated_score : null,
           typeof item.breaking_score === "number" ? item.breaking_score : null,
           nonEmptyText(item.breaking_label),
           nonEmptyText(item.breaking_reason),
           typeof item.breaking_confidence === "number" ? item.breaking_confidence : null,
           jsonObjectText(item.breaking_dimensions),
+          jsonObjectText(item.breaking_adversarial_flags),
           nonEmptyText(item.breaking_score_version),
           nonEmptyText(item.target_timezone),
           nonEmptyText(item.published_at_local),
@@ -260,9 +273,11 @@ export async function importEventsToD1(
            image_urls, tags, issue_tags, related_tags, region_tags, entities,
            language, pipeline_stage,
            value_label, value_score, china_relevance_label, classification
-           , breaking_score, breaking_label, breaking_reason, breaking_confidence,
-           breaking_dimensions, breaking_score_version, target_timezone, published_at_local
-         ) VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)
+           , breaking_raw_score, breaking_percentile, breaking_calibrated_score,
+           breaking_score, breaking_label, breaking_reason, breaking_confidence,
+           breaking_dimensions, breaking_adversarial_flags, breaking_score_version,
+           target_timezone, published_at_local
+         ) VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)
          ON CONFLICT(event_id) DO UPDATE SET
            target_id=excluded.target_id,
            target_label=excluded.target_label,
@@ -297,11 +312,21 @@ export async function importEventsToD1(
              events.china_relevance_label
            ),
            classification=COALESCE(NULLIF(excluded.classification, '{}'), events.classification),
+           breaking_raw_score=COALESCE(excluded.breaking_raw_score, events.breaking_raw_score),
+           breaking_percentile=COALESCE(excluded.breaking_percentile, events.breaking_percentile),
+           breaking_calibrated_score=COALESCE(
+             excluded.breaking_calibrated_score,
+             events.breaking_calibrated_score
+           ),
            breaking_score=COALESCE(excluded.breaking_score, events.breaking_score),
            breaking_label=COALESCE(NULLIF(excluded.breaking_label, ''), events.breaking_label),
            breaking_reason=COALESCE(NULLIF(excluded.breaking_reason, ''), events.breaking_reason),
            breaking_confidence=COALESCE(excluded.breaking_confidence, events.breaking_confidence),
            breaking_dimensions=COALESCE(NULLIF(excluded.breaking_dimensions, '{}'), events.breaking_dimensions),
+           breaking_adversarial_flags=COALESCE(
+             NULLIF(excluded.breaking_adversarial_flags, '{}'),
+             events.breaking_adversarial_flags
+           ),
            breaking_score_version=COALESCE(
              NULLIF(excluded.breaking_score_version, ''),
              events.breaking_score_version
@@ -342,11 +367,15 @@ export async function importEventsToD1(
         item.value_score ?? null,
         item.china_relevance_label || "未知",
         jsonText(item.classification, {}),
+        item.breaking_raw_score ?? item.breaking_score ?? item.value_score ?? null,
+        item.breaking_percentile ?? item.breaking_score ?? item.value_score ?? null,
+        item.breaking_calibrated_score ?? item.breaking_score ?? item.value_score ?? null,
         item.breaking_score ?? item.value_score ?? null,
         item.breaking_label || null,
         item.breaking_reason || null,
         item.breaking_confidence ?? null,
         jsonText(item.breaking_dimensions, {}),
+        jsonText(item.breaking_adversarial_flags, {}),
         item.breaking_score_version || null,
         item.target_timezone || "UTC",
         item.published_at_local || null,
@@ -407,6 +436,7 @@ export async function handleImport(
     const body = await importEventsToD1(db, events);
     if (body.imported > 0 || body.updated > 0) {
       try {
+        await refreshBreakingScoreStats(db);
         await refreshPublicReadSnapshots(db);
       } catch (error) {
         console.warn("public snapshot refresh after import failed:", error);