chore: add ClickHouse archive scripts for citation rank data

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-07-03 20:55:33 +08:00
parent 15a820712e
commit 4dece0f26a
3 changed files with 654 additions and 0 deletions
@@ -0,0 +1,39 @@
+DROP VIEW IF EXISTS geo_monitoring_archive.daily_model_site_citation_rank;
+
+CREATE VIEW geo_monitoring_archive.daily_model_site_citation_rank AS
+SELECT
+  business_date,
+  model,
+  ai_platform_id,
+  site,
+  hosts,
+  citation_count,
+  cited_run_count,
+  unique_url_count,
+  brand_count,
+  rank() OVER (
+    PARTITION BY business_date, model
+    ORDER BY citation_count DESC, cited_run_count DESC, unique_url_count DESC, site ASC
+  ) AS site_rank
+FROM
+(
+  SELECT
+    cf.business_date AS business_date,
+    ifNull(nullIf(r.provider_model, ''), cf.ai_platform_id) AS model,
+    cf.ai_platform_id AS ai_platform_id,
+    cutToFirstSignificantSubdomain(cf.host) AS site,
+    groupUniqArray(cf.host) AS hosts,
+    count() AS citation_count,
+    uniqExact(cf.run_id) AS cited_run_count,
+    uniqExact(cf.normalized_url) AS unique_url_count,
+    uniqExact(cf.brand_id) AS brand_count
+  FROM geo_monitoring_archive.monitoring_citation_facts AS cf
+  LEFT JOIN geo_monitoring_archive.question_monitor_runs AS r
+    ON cf.tenant_id = r.tenant_id
+   AND cf.run_id = r.id
+  GROUP BY
+    cf.business_date,
+    model,
+    cf.ai_platform_id,
+    site
+);